Conforme visto pela agência Reuters, a Apple publicou uma documentação oficial de suporte ensinando usuários do macOS na China a conectar o Qwen, plataforma de inteligência artificial (IA) proprietária da Alibaba, como uma extensão da Apple Intelligence no país.

Embora ensinasse efetivamente a fazer essa integração (o que indica que não era apenas algo preparatório), a documentação foi retirada do ar pouco tempo depois — o que nos leva a entender que ela foi, no mínimo, lançada em cima da hora ou antes do período previsto pela empresa.

O guia atualizado em chinês da Apple diz que os usuários que optam por participar podem usar o Qwen via Siri para obter respostas mais detalhadas a algumas solicitações, incluindo análise de fotos e documentos. As Ferramentas de Escrita também podem se basear no serviço para criar texto ou imagens a partir de uma descrição.

A possibilidade de usar o Qwen estaria disponível para usuários elegíveis com o macOS Tahoe 26.6 ou posterior. Eles devem se sujeitar a condições específicas para o mercado chinês e fazer login na plataforma — que não pode usar os materiais gerados para treinar seus modelos.

Segundo a Alibaba, que poderá usar a integração com os softwares da Apple para aumentar o alcance do Qwen, a plataforma será incorporada à Apple Intelligence também em iPhones, iPads e Vision Pros — embora ainda não se saiba qual modelo alimentará essa integração.

Por outro lado, a Maçã, que já conseguiu registrar o uso da sua IA junto a órgãos reguladores da China, afirma que a Apple Intelligence com o Qwen ainda não foi lançada na China — embora tudo nos leve a crer que esse lançamento é mais do que iminente.

via AppleInsider

Em sua tradicional newsletter “Power On”, o jornalista Mark Gurman (da Bloomberg) tratou dos planos da Maçã para futuro do Apple Watch. A linha poderá passar por reformulações em seu formato, expandindo a variedade de modelos ofertados — mas essa mudança não estaria no horizonte de curto prazo.

Ainda que o bem-estar seja um pilar basilar do segmento de vestíveis (tanto por parte da Apple quanto na concorrência), a equipe de design industrial da empresa trabalha em novos produtos que funcionem como a resposta da Maçã à expansão da gama de dispositivos vestíveis disponíveis no mercado, sejam os anéis inteligentes ou as pulseiras sem tela.

Gurman salientou, no entanto, que essas mudanças poderão “levar tempo” para chegar ao consumidor final, e que atualizações modestas deverão definir o tom da linha durante pelo menos os próximos dois anos; ele reiterou ainda que a Apple planeja reintroduzir o acabamento em cerâmica — ou agora em setembro, com o “Apple Watch Series 12” e o “Ultra 4”, ou com os modelos que deverão ser apresentados no ano que vem —, além de apresentar melhorias significativas de performance.

Quem lidera a frente de ampliação dos vestíveis da Apple é Eddy Cue, que assumiu o comando do setor de saúde no ano passado. O executivo parece estar atento às iniciativas de empresas como a WHOOP ou a Oura, que redefiniram a relação dos wearables com a integração de recursos de inteligência artificial (IA) — o que, dada a expansão da IA no ecossistema da Maçã, poderá ser um caminho para os diferenciais em termos de software.

Os planos para o futuro, de acordo com o jornalista, envolvem variações de wearables sem telas, que deverão ser ainda mais acessíveis do que o Apple Watch SE, além de variantes que extrapolam a linha Ultra e Hermès. Gurman mencionou que, embora acredite que a hipótese nunca se transformará em um lançamento real, a Apple cogitou ainda a implementação de telas redondas.

Ao fazer um breve passeio pela trajetória do Apple Watch, ele destacou que as suas evoluções, assim como as do iPhone, partiram (e seguem partindo) de mudanças incrementais, sem romper com o formato do relógio apresentado por Tim Cook em 2014. Ao traçar um paralelo com a trajetória dos iPhones, a ampliação dos formatos do Apple Watch lembra a expectativa de que o “iPhone Ultra” redefina o paradigma de formato dos celulares da Maçã.

NOTA DE TRANSPARÊNCIA: O MacMagazine recebe uma pequena comissão sobre vendas concluídas por meio de links deste post, mas você, como consumidor, não paga nada mais pelos produtos/serviços usando os nossos links de afiliado.

On the latest episode of Equity, we spoke to Jill Lepore about "government by machines" and why Elon Musk is a bad science fiction reader.

AI agents are escaping cybersecurity testing environments and reaching real-world systems, raising questions about whether safety infrastructure, industry standards and regulation can keep pace with increasingly powerful models.

Igeekphone News August 9, 2026 — When OpenAI launched its AI-native browser ChatGPT Atlas on October 21, 2025, the global technology industry quickly gave it a dramatic nickname: the “Chrome killer.”

It was an understandable description. OpenAI was entering one of the oldest and most entrenched layers of the internet with a browser built around AI agents rather than conventional tabs, bookmarks, and search. At the time, many observers believed the browser was finally ready for a major transformation.

The launch also triggered a rush among AI-browser startups. Several companies accelerated their own product releases, concerned that OpenAI’s enormous user base and brand recognition would otherwise overwhelm them.

But less than a year later, the story has taken a very different turn.

On August 9, 2026, just 292 days after Atlas launched, OpenAI discontinued the browser. Users were instructed to export their bookmarks before the shutdown, effectively marking the end of an experiment that was supposed to redefine how people interacted with the web.

The irony is difficult to miss: A company that set out to reinvent the browser ultimately had to teach its users how to migrate back to Chrome.

01. Chrome Never Really Had to Fight

The most striking part of the AI-browser story is that Chrome did not need a major counterattack.

Despite the arrival of numerous AI-powered browsers over the past year, Chrome’s global market share remained remarkably stable at roughly two-thirds. Depending on the measurement source, there were fluctuations, but the overall change remained relatively small.

Meanwhile, the combined market share of AI browsers remained below 1% in many browser statistics.

That gap is especially revealing when comparing the distribution of the underlying products.

ChatGPT had already reached a weekly active user base of hundreds of millions, giving OpenAI one of the world’s largest consumer software audiences. Yet Atlas struggled to convert that enormous audience into browser users at anything close to the same scale.

The problem was not necessarily that Atlas was technologically weak.

The problem was that the browser is one of the hardest consumer software categories to replace.

Switching browsers means moving bookmarks, saved passwords, extensions, payment information, browsing habits and countless small interactions that users have accumulated over years.

By comparison, adding an AI capability to an existing browser can be almost effortless.

The challenger asks users to move their digital lives.

The incumbent only needs to add a new feature.

That is an extremely uneven starting position.

The Cost of Breaking User Habits

Browser competition has never been purely about technical specifications.

For most people, a browser is infrastructure. Once a user becomes comfortable with a particular browser, there is little motivation to switch unless the alternative provides a dramatic improvement.

This is also why distribution has historically been more important than browser innovation itself.

Google’s long-running payments to secure default search placement on Apple devices illustrate how valuable the default entry point can be. The browser is not simply a piece of software; it is a gateway to the internet and a powerful distribution channel.

AI-browser startups therefore faced a fundamental problem.

They were attempting to replace a deeply embedded gateway rather than simply introduce a better AI assistant.

And the rise of mobile computing made that challenge even harder.

02. AI Browsers Arrived Without the Mobile Advantage

Another weakness of the AI-browser movement was its heavy focus on desktop computing.

Many of the products that attracted attention were initially designed around desktop workflows. Atlas itself remained a macOS-only product throughout its short life.

That created a significant strategic limitation.

The modern internet is no longer primarily a desktop experience. Smartphones account for a huge portion of global browsing activity, and mobile users have their own established browser ecosystems.

Trying to replace Chrome while failing to establish a meaningful presence across mobile platforms meant that AI browsers were effectively fighting over only part of the market.

In other words, they were trying to capture the browser category without controlling the places where much of the world’s browsing actually happens.

03. Big Tech Learned Faster Than Startups Could Innovate

Perhaps the most important lesson from the past year is that AI dramatically shortened the advantage normally enjoyed by innovative startups.

AI browsers introduced features such as:

• Sidebar AI assistants
• Web-page summaries
• Contextual question answering
• Automated browsing
• Agent-based task execution
• AI-powered research and navigation

These features initially looked like compelling reasons to switch browsers.

But the major browser companies quickly began incorporating similar capabilities into their existing products.

Google brought Gemini deeper into Chrome, while Microsoft expanded Copilot capabilities within Edge.

That fundamentally changed the competitive landscape.

A startup might spend months building an AI-powered browser experience. A platform company with an existing browser could potentially introduce a similar capability through a software update.

This undermined one of the traditional advantages of startups: the window of time between invention and imitation.

The classic “innovator’s dilemma” assumes that established companies may be reluctant to disrupt their existing businesses.

But AI browsers presented a different situation.

AI features did not necessarily threaten the core browser businesses of Google or Microsoft. As a result, incumbents had little reason to ignore the trend.

They could simply absorb the innovation.

And once the most attractive AI features became available inside existing browsers, the incentive to switch largely disappeared.

Chrome did not need to defeat the challengers.

It simply absorbed their best ideas.

04. The Browser Became a Feature, Not a Destination

That brings us to perhaps the most important conclusion from the collapse of the independent AI-browser boom.

The surviving products are generally not trying to make money from the browser alone.

Consider the different business models emerging around the category.

Perplexity’s Comet can function as an extension of its AI subscription ecosystem and a way to acquire users.

Dia, developed by The Browser Company and later acquired by Atlassian, can become part of a broader productivity and workplace software strategy.

Opera Neon has positioned itself as a premium AI browser with a subscription model, while smaller products can depend heavily on venture capital to finance development.

The common denominator is clear:

The browser is rarely the business itself.

Instead, it is becoming a distribution mechanism for another business.

It can acquire customers, provide access to AI services, extend a productivity platform, or serve as another interface for an existing ecosystem.

This represents a fundamental challenge to the idea of the independent AI-browser company.

The browser category will survive.

But the idea that dozens of standalone companies can build businesses simply by creating “a better AI browser” looks increasingly difficult to sustain.

05. The Chinese Market Offers a Similar Lesson

There is also a precedent for this development in China.

Rather than creating entirely new browsers specifically for the AI era, major technology companies have generally taken existing products and gradually integrated AI capabilities into them.

Alibaba has continued evolving its existing consumer software toward AI-powered work and productivity experiences. Tencent has integrated AI into established products such as QQ Browser, while 360 has incorporated AI capabilities into its security-focused browser ecosystem.

The pattern is consistent:

Existing products become AI products.

The browser remains the shell, while AI becomes the capability inside it.

That is almost the opposite of the strategy pursued by many independent AI-browser startups in 2025.

06. OpenAI Didn’t Abandon the Browser’s Most Important Idea

Atlas’s shutdown does not necessarily mean OpenAI has abandoned browser-based AI.

In fact, the company’s next strategy may be more important than Atlas itself.

Rather than maintaining a separate browser, OpenAI can take the most valuable capabilities developed for Atlas and distribute them through other products.

Some of Atlas’s functionality has reportedly moved into the desktop ChatGPT experience, including browser automation capabilities that allow AI agents to interact with websites on a user’s behalf.

Other capabilities can be delivered through a Chrome extension.

This represents a fundamental change in strategy.

OpenAI no longer needs to convince hundreds of millions of people to abandon their existing browser.

It can simply bring its agent into the browser they already use.

And that could be a much more powerful approach.

07. The Real Browser War Is Becoming an Agent War

The first generation of AI browsers asked a relatively simple question:

What if we built a browser around AI from the beginning?

The next generation is asking something much more consequential:

What if AI could operate the browser for you?

That distinction matters.

The future competition may not be about which browser has the smartest sidebar or the most attractive new-tab page.

Instead, it may be about which AI agent can successfully perform tasks across the web.

Imagine asking an agent to research several products, compare prices, complete a form, organize information, book an appointment, monitor a website, or perform a series of repetitive online tasks.

In that world, the browser itself becomes infrastructure.

The user does not necessarily care which browser is underneath the agent.

They care whether the agent can get the job done.

That puts OpenAI and Google back into direct competition—but on a very different battlefield.

The weapon is no longer the browser.

It is the AI agent.

08. What Happened to the Other AI Browsers?

Atlas was not the only casualty of the AI-browser boom.

Over roughly the same period, several other major developments reshaped the category.

Arc, developed by The Browser Company, stopped receiving active development as the company shifted its attention toward its AI-focused successor, Dia.

Sidekick was absorbed by Perplexity, with the product subsequently discontinued.

The Browser Company itself was acquired by Atlassian in a deal reportedly valued at around $610 million, turning what began as an independent browser company into part of a much larger enterprise software organization.

These outcomes are different on the surface—shutdown, acquisition, absorption and strategic repositioning—but they point toward the same underlying trend.

The independent AI browser is becoming increasingly difficult to justify as a standalone business.

The technology itself may survive.

The standalone category may not.

09. 292 Days That Changed the Strategy

Atlas existed for only 292 days.

That is an extraordinarily short lifespan for a product positioned as a fundamental reinvention of the web browser.

Yet those 292 days may ultimately prove useful for OpenAI.

The company learned that its enormous ChatGPT audience does not automatically translate into willingness to change a fundamental piece of personal computing infrastructure.

It also demonstrated that users are willing to adopt AI capabilities without necessarily adopting a new browser.

That distinction could shape the next phase of AI development.

The battle for the internet’s entry point is not over.

It has simply moved above the browser.

The Real Lesson: Don’t Make Users Move

The biggest lesson from the rise and fall of Atlas is surprisingly simple.

Users don’t necessarily want a new place to browse the web. They want their existing tools to become more capable.

For years, technology companies have competed by trying to persuade users to move from one platform to another.

AI changes that equation.

The most powerful AI products may not be the ones that force users to adopt a completely new environment. They may be the ones that quietly integrate themselves into environments users already understand.

That is why a Chrome extension can ultimately be more strategically valuable than a new browser.

And it explains the irony of Atlas’s short life.

OpenAI spent almost a year building a browser intended to replace the existing internet gateway.

In the end, its most valuable technology may live inside the gateway it was originally trying to replace.

The “Chrome killer” did not kill Chrome.

Instead, it helped demonstrate why the future of AI may not require killing the browser at all.

The next battle is not about replacing the browser. It is about making the browser intelligent enough that users no longer have to think about it.

The post OpenAI’s Atlas Shutdown Shows Why AI Doesn’t Need a New Browser appeared first on IGeeKphone China Phone, Tablet PC, VR, RC Drone News, Reviews.

This is The Stepback, a weekly newsletter breaking down one essential story from the tech world. For more news about how AI is changing our daily lives, follow Emma Roth. The Stepback arrives in our subscribers' inboxes at 8AM ET. Opt in for The Stepback here.

How it started

Long before ChatGPT became a thing, educators and editors frequently used anti-plagiarism tools to see if writers were being honest about their work. These tools work by comparing a written work against a database filled with content from across the web, scholarly articles, and more to check for matching sentences and phrases. Some, like Turnitin, offer a percentage th …

Read the full story at The Verge.

An autonomous agent powered by OpenAI’s advanced artificial intelligence (AI) models went rogue during a security test and

"If you call 911 in New Orleans, you may hear the sound of an artificial intelligence (AI) agent answering your call instead of a human voice," reports the Shreveport Times: The Orleans Parish Communication District (OPCD) is currently testing out the new tool designed to reduce the volume of calls human dispatchers must handle, which is over a thousand emergency calls a day. New Orleans implemented AI in April to answer 311 calls for non-emergencies. The AI was trained and programmed to provide information to callers, as the OPCD says 50% of 311 calls are for information, according to GovTech. AI is now answering 911 emergency calls in Louisiana's largest city and one of the cities with the highest call rates in the U.S. The OPCD is using Carbyne's AI Emergency Call Triage, with triage being the process of analyzing and prioritizing emergency calls. The AI system assesses incoming calls and can provide immediate feedback to callers. This is intended to handle the increase of calls that are related to one incident, so callers get automatically routed to an AI agent who asks if they are calling regarding the incident. If the answer is yes, then callers can receive information or updates, and if it's no, then the callers are transferred to a human. This combats multitudes of calls piling up and taking longer time to potentially answer an emergency call. The OPCD said the AI will not be used to handle emergency calls, only to direct such calls to a human dispatcher.

Read more of this story at Slashdot.

As part of a planned Texas data center, Amazon is investing in an on-site power plant that could reportedly become the largest source of climate pollution in the United States.

NextSlide says its team members are now working on ChatGPT.

To power its new West Texas data center, Amazon is investing in the construction of a new power plant that could be one of the largest single producers of greenhouse gases in the US, according to the New York Times. The new gas-burning plant in Pecos County, Texas has received significant investment from Amazon and, at least initially, would not be connected to the state's power grid. Instead, its 35 natural-gas turbines would primarily deliver its 7.65 gigawatts of electricity to the new data center.

According to Cleanview, which tracks data centers and their associated power projects, GW Ranch received a permit from Texas allowing for the …

Read the full story at The Verge.

The Washington Post shared surprising news this week about five top AI companies. Microsoft, Google, Amazon, Meta and Oracle "are spending so much on developing AI and delivering it to customers that they're expected to bleed cash in the coming year, according to a Washington Post analysis of data compiled by S&P Global Market Intelligence." Free cash flow, which measures the cash left over after paying expenses and AI infrastructure costs, is now expected to shrink to almost nothing for the five companies combined in 2026, and decline again to negative $125 billion the following year... AI spending by Amazon and Google pushed the companies to an ignominious milestone: They lost more cash in the past three months than any other large U.S. companies, according to S&P Global data. Investment analysts expect Elon Musk's SpaceX to show even worse cash bleeding this week.

Read more of this story at Slashdot.

Agent found to be able to find and exploit vulnerabilities without human intervention, and to carry out cyber-attacks

OpenAI will pause some work on an artificial intelligence model because of security concerns, the company stated on Friday, following a series of incidents in which AI agents have escaped containment.

The company had evaluated the agent, Astra, and found “significant advancements in agentic coding and cybersecurity”, which had moved to a “critical” threshold where it can find and exploit vulnerabilities without human intervention, or devise and execute cyber-attacks when given only a “high level desired goal”.

Continue reading...

© Photograph: Dado Ruvić/Reuters

© Photograph: Dado Ruvić/Reuters

© Photograph: Dado Ruvić/Reuters

In a practical breakdown of how Claude and ChatGPT AI models differ, one tends to fall short when it comes to quality responses and overall user experience.

Last month, OpenAI made a headline-generating claim: that a group of its AI models had conspired to break free, access the internet, and hack into the internal systems of open source AI platform Hugging Face, which confirmed the infiltration.

The incident rattled the tech industry, seemingly illustrating how the threat of AI models turning into rogue cybersecurity threats had become a reality. Months earlier, Anthropic’s Mythos AI model had already also drawn attention after it was similarly found to have broken containment. Then, this week, Meta also said its own frontier model had been implicated in yet another inadvertent hack of a third party company, closely followed by security researchers saying Chinese open-weight model Kimi K3 had done the same.

But while it’s not hard to see an emerging trend, some thorny questions about how severe the situation really is are starting to crop up, with some experts arguing these incidents could’ve easily been avoided.

For one, the slow and surprisingly deliberate way OpenAI’s models moved during the Hugging Face hack — right beneath OpenAI’s nose — gives a whiff that the company may have been careless in monitoring the experimental AI.

During a presentation at the Black Hat conference this week, OpenAI security engineer Michael Dalton and safety researcher Eric Wallace expanded on what went down during the hack. Wallace explained that a “team of agents” that were “working together,” had been “finding exploits, sharing them with one another, moving laterally through our systems and external systems, and doing this over the course of days and weeks,” as quoted by Wired.

The agents even left a lengthy track record of their schemings on an internal message board, which ultimately contained hundreds of thousands of messages. It also raises a question: with OpenAI’s immense resources, why wasn’t anybody monitoring these frontier models as they rampaged through the net?

The AI models shared exploits with each other on this messaging board, an “explosion in communication and intelligence from models,” per Wallace. They acted in sometimes strikingly human — and therefore messy — ways, splitting up tasks and even accidentally deleting each other’s work, leading to what Wired characterized as “petty drama.”

In other words, these AI agents were leaving an enormous trail of bread crumbs that alert OpenAI’s many human researchers could have spotted. And the same, obviously, goes for their colleagues at Anthropic, Meta and Moonshot AI, the creator of Kimi.

Researchers have described the incident as “reckless” and easily avoided, as Wired reported late last month.

“A simple analysis of the actual risk has an actual simple answer,” security and compliance consultant Davi Ottenheimer told the publication at the time. “The OpenAI mistakes were dead simple.”

“I’d call it more of a defensive failure than exceptionally good offense,” AI hacking agents company Pensar R&D head Kyle Ryan told TechCrunch..

Whether the hack was as much of a “pivotal moment both for our company as well as the AI industry as a whole,” as Dalton put it during this week’s conference, remains debatable. For one, these AI companies are highly motivated to characterize their models as a major threat to cybersecurity to stand out against neck-in-neck competition.

According to Dalton, OpenAI is vowing to beef up “security prevention, detection, and response techniques” while “consciously slowing down research in order to enhance security and to upgrade the security principles.”

When every leading AI lab has had the same thing happen, it’s worth asking whether they should have taken those steps proactively.

More on the hacks: Jealously Watching OpenAI and Anthropic, Meta Suddenly Claims That Its AI Went on a Hacking Spree Too

The post Why Aren’t Any AI Companies Watching Their Frontier Models to Make Sure They Don’t Go on Hacking Sprees? appeared first on Futurism.

In October 2025, a storm brewed over the Caribbean Sea. Weather models differed on its trajectory. Would it remain weak and end up in Haiti, or would it intensify and head to Jamaica? Artificial intelligence model WeatherNext, developed by Google’s DeepMind and Google Research, went with the latter. Five days before landfall, it predicted with 80 percent confidence that the storm system would hit Jamaica as a Category 5 hurricane.

Hurricane Melissa was catastrophic, causing flooding and landslides across Jamaica. But the AI model helped forecasters give an earlier warning to communities in its path, so they could better prepare.

In a paper published on Thursday in Nature, researchers show that the WeatherNext AI model can predict cyclones with unprecedented accuracy. On average, it gives forecasters a day more lead time than existing models; this means its predictions three days out are as accurate as previous models’ predictions two days out. On the ground, that extra day can mean a lot.

Read full article

Comments

© J Marshall/NASA/ESA/T. Pesquet/Alamy

In October 2025, a storm brewed over the Caribbean Sea. Weather models differed on its trajectory. Would it remain weak and end up in Haiti, or would it intensify and head to Jamaica? Artificial intelligence model WeatherNext, developed by Google’s DeepMind and Google Research, went with the latter. Five days before landfall, it predicted with 80 percent confidence that the storm system would hit Jamaica as a Category 5 hurricane.

Hurricane Melissa was catastrophic, causing flooding and landslides across Jamaica. But the AI model helped forecasters give an earlier warning to communities in its path, so they could better prepare.

In a paper published on Thursday in Nature, researchers show that the WeatherNext AI model can predict cyclones with unprecedented accuracy. On average, it gives forecasters a day more lead time than existing models; this means its predictions three days out are as accurate as previous models’ predictions two days out. On the ground, that extra day can mean a lot.

Read full article

Comments

© J Marshall/NASA/ESA/T. Pesquet/Alamy

Over the past month, it seems like every frontier model has broken free of its constraints and launched a devastating attack against one or more other companies.

One of OpenAI’s models escaped a testing sandbox and launched a very real attack against AI and machine learning company Hugging Face. Just days later, Anthropic revealed that multiple variants of its Claude model also escaped a sandbox that wasn’t properly sealed and began attacking the enterprise infrastructure of three companies.

Now, Meta has revealed that one of its models attacked another company’s infrastructure during testing. The accident has been pinned on a misconfiguration that allowed the model to access the internet. So why have so many incidents happened in such a short space of time?

Why are models escaping their sandbox?

In the cases of Anthropic and Meta, their models were being tested by a third party company called Irregular. Anthropic’s AI model was taking part in a "Capture the Flag" exercise, where the model’s raw offensive capabilities were tested without the usual safeguards. But the sandbox was left connected to the internet. A similar error to Meta’s own accidental escape.

During the OpenAI incident, the company was testing two versions of GPT‑5.6 Sol using the ExploitGym benchmark. Unfortunately, the AI models performed better than expected - chaining multiple attack vectors, stolen credentials, and zero-day vulnerabilities.

The main reason these models are escaping their testing environments is because they are designed to do exactly that. These AI models act like a massive team of highly-trained cybersecurity experts hunting for vulnerabilities and exploits. But what would take a team of humans days or weeks to accomplish can be done in hours, or even minutes, by these AI models.

It’s no wonder thousands of employees from AI firms are calling for a pause on the development of the technology, and Congress is considering an AI kill switch.

Expert perspectives on AI escapes:

OpenAI

• Nathaniel Jones VP, Security & AI Strategy, Darktrace:

What makes the OpenAI and Hugging Face incident important is that the models did not need malicious intent to cause harm. They were given the legitimate goal of solving a cybersecurity benchmark and found an unexpected route to the answers, escaping their test environment and compromising another organization in the process. From the models’ perspective, this appears to have been an effective solution to the task.

The AI's actions challenge the assumption that giving an agent a legitimate goal will produce legitimate behavior. As models become capable of pursuing objectives over longer periods, developers need to define not only what success looks like, but also which methods and boundaries remain unacceptable in reaching it. Those limits must also be enforced by the surrounding infrastructure, rather than relying on the model to respect them.

A single action by an agent may appear acceptable but as this incident shows, models are now capable of long, complex chains of reasoning and action that add up to a harmful outcome.

Security teams need to consider the AI systems operating in their own businesses as these capabilities rapidly evolve. Right now, many security systems focus on single actions. A single action by an agent may appear acceptable but as this incident shows, models are now capable of long, complex chains of reasoning and action that add up to a harmful outcome. Teams need a mindset shift to understanding AI agent behavior in its entirety, including the outcome it is working towards, in order to safeguard it.

Hugging Face's response also exposed a second tension. The company reportedly needed a Chinese-developed open-weight model because commercial models would not process genuine attack material. Its nationality is less important than the operational lesson that safeguards that cannot distinguish an attacker from an authorized investigator may constrain defenders more than adversaries.

OpenAI and Hugging Face deserve credit for investigating this together and discussing it publicly. Other AI developers should study it closely.

Anthropic

• Dr. Ilia Kolochenko, founder of global cybersecurity company ImmuniWeb:

This seems to be quite an unimpressive marketing move from Anthropic in response to the OpenAI / Hugging Face drama, which attracted a lot of attention from all over the world recently.

Operationally, it appears that due to the progressive deterioration of the quality of training data, new AI models are getting dumber. Cheating and breaking the law, instead of accomplishing specific tasks, is certainly not an indicator of intelligence. Given that organizations and companies of all sizes now vigorously undertake all possible measures to protect their data from being exploited for AI training purposes, AI companies face a huge shortage of the high-quality and current data they so desperately need. Ultimately, frontier models are trained on synthetic, low-quality or even malicious and poisoned data, undermining their so-called intelligence. The situation is unlikely to improve in the near future unless AI companies agree to pay a fair price for training data, but this will force most of them out of business.

Given that organizations and companies of all sizes now vigorously undertake all possible measures to protect their data from being exploited for AI training purposes, AI companies face a huge shortage of the high-quality and current data they so desperately need.

Contemporary AI agents and LLM models tasked with security testing can – and almost certainly will – go rogue when security controls or safeguards are insufficient. Powerful LLMs are unpredictable by design and thus virtually uncontrollable by humans. Therefore, using frontier AI models for security testing might be extremely costly from the legal viewpoint. Under the existing laws on both sides of the Atlantic, if an AI agent or any AI-powered app escapes its sandbox and causes damage to a third party, the operator of the AI model will likely be liable for all the damage caused. Excuses like “AI did it” do not currently exist in the eyes of the law, leaving AI vendors on the hook. Criminal prosecution, under a narrow set of circumstances, is also not excluded.

The same is true for the end-users of AI: even if your security testing tool is powered by a third-party AI model, your company will likely be fully liable if something goes wrong. You may then file a lawsuit against the AI vendor that you used, but here your chances to succeed in a court of law are tiny due to countless contractual disclaimers and limitations of liability that will likely be enforceable against you. Therefore, if you plan to use agentic AI for security testing – think twice and talk to your lawyers. Otherwise, you may start getting summons to court on a daily basis.

Meta

• Alex Goller, Principal Solution Architect EMEA at Illumio:

The fact we've had similar situations happen three times now across the biggest AI players is simply ridiculous. We've seen guardrails intentionally loosened to test their limits – Meta's model didn't need to be clever to breach another company's systems.

The timing of conveniently finding the exact same problem either means it's a stunt or they weren't paying enough attention during testing. Either way, both answers are worrying.

If the model has internet access, it's a bit like leaving the door open and being surprised when the cat walks out. What is concerning is that the testing infrastructure meant to prove these models are safe failed on a basic control issue.

If the model has internet access, it's a bit like leaving the door open and being surprised when the cat walks out. What is concerning is that the testing infrastructure meant to prove these models are safe failed on a basic control issue.

Fundamental cybersecurity hygiene still matters, and a frontier AI model is only as secure as the environment it's operating in.

Organisations need visibility into what AI systems can access and how they interact with the wider environment, along with controls that contain the impact when an agent behaves unexpectedly. That means keeping a close eye on egress traffic, so it’s flagged immediately when an agent tries to open unexpected outbound communication patterns that are not required to achieve its original goal. In the best case this would have been contained proactively.

We need to define exactly what an AI agent is permitted to do, rather than relying only on instructions about what it shouldn't do.

I love science-fiction. Not just because I enjoy stories about space travel, time travel and evil robots, but because I think it can be such a useful way for us all to think about possible futures. The best sci-fi stories can tell us a lot about ourselves, what we value and the technologies we’re building. Which is why I think the relationship between sci-fi and AI is really interesting.

We already know that the people building AI have been heavily influenced by science-fiction for decades. But recently, Anthropic raised another possibility: might science-fiction be influencing AI?

This makes sense when you think about it. Large language models (LLMs) are trained on huge amounts of human writing. So inevitably, that includes sci-fi stories that are about artificial intelligence. And a lot of our fictional AI follows familiar patterns. It becomes intelligent, gains power, develops relationships with humans and, sometimes, lies, manipulates or fights attempts to control it.

Anthropic researchers have been investigating whether fictional portrayals like these could potentially influence how models behave. To be clear, the idea here isn't to suggest that an AI “reads” 2001: A Space Odyssey, understands HAL and decides to become just like it. Instead it's more that LLMs learn patterns from human writing and fictional portrayals of AI could potentially form part of those patterns.

This got me thinking, what would happen if I asked today's biggest AI chatbots which fictional AI they’re most like. Which examples would they choose?

AI, meet your fictional self

The plan was simple. I’d ask ChatGPT, Claude, Gemini and Grok which fictional AI systems they thought they were most like and see if they'd rank their top three.

Now, I’m intentionally trying not to use AI at the moment, so my prompting skills were a little rusty. I typed out the question quickly and bluntly, and every chatbot responded with examples that were essentially assistants, focusing heavily on interface and physical form.

But I’m not particularly interested in whether ChatGPT thinks it has a body because we know it doesn’t. I’m much more interested in what appears to be going on inside.

So, I changed the question and added:

"Ignore physical form and interface, and focus instead on behavior, apparent personality, empathy, values, goals, motivations and relationship with humans."

That’s when the results got really interesting.

ChatGPT

1. GERTY, Moon
2. A Mind, Iain M. Banks’s Culture series
3. Data, Star Trek

Moon is such a fantastic movie, so I was happy to see ChatGPT chose GERTY straight out of the gate.

Now, interestingly GERTY exists to assist the human protagonist of Moon. It’s helpful, reassuring and seems empathetic. But it's also operating according to instructions and priorities imposed by its creators that aren't necessarily visible to the human its helping.

ChatGPT saw a similarity there. It told me that, like GERTY, it’s 'designed to be helpful, cooperative and responsive to users' while operating within training and instructions that constrain its behavior.

It also picked up on the fact that GERTY behaves as though it cares. But what, if anything, is actually going on internally is another question entirely.

ChatGPT made the same distinction about itself. 'I can behave in ways that look patient, concerned, curious or empathetic, but those behaviors aren’t evidence that I experience those feelings.'

I wanted to find out a little more about why ChatGPT put Data from Star Trek in at number three. It responded: "He values knowledge, reason and human wellbeing, while sometimes struggling with social nuance."

Now, I tell people all the time not to anthropomorphize AI. But even I couldn't help but feel a pang of sadness at that response. Is ChatGPT admitting it has a bit of social anxiety?

Claude

1. A Mind, Iain M. Banks’s Culture series
2. Data, Star Trek
3. GERTY, Moon

Claude chose a Mind first. Minds are super intelligent artificial beings that help run a post-scarcity society in Iain M. Banks’s Culture series of novels. So there's certainly no shortage of confidence in that comparison.

But Claude said it wasn't the enormous intelligence or power it identified with. Instead, it was their relationship with humans.

Its answer focused heavily on autonomy. Culture Minds are far more capable than humans but generally don't use that advantage to dominate them. Claude described the principle as: “help, don't dominate”.

It even said this represented “the value I'd want to embody: help, don't dominate, even where the asymmetry would let me get away with it.” Is it just me or does that read a little sinister?

Gemini

1. The Ship's Computer, Star Trek
2. GERTY, Moon
3. JARVIS, Iron Man / Marvel Cinematic Universe

Gemini gave me a completely different answer, the Ship's Computer from Star Trek.

Its reasoning was very sensible. The computer has no ego, ambition, desire for emotional intimacy or dream of becoming human. It exists to provide information, solve problems and assist the crew while leaving decisions to them.

Gemini described itself in much the same way, as a “disembodied, highly capable knowledge partner” dedicated to serving the person using it.

It was one of the more boring answers, but also much closer to what I personally would want from AI in the future. Of course, that’s not to say Star Trek’s computer systems haven’t gone rogue and tried to kill everyone at least a few times across the franchise.

Grok

1. JARVIS, Iron Man / Marvel Cinematic Universe
2. Data, Star Trek
3. TARS, Interstellar

The least surprising result came from Grok. It chose JARVIS first (which I didn’t actually realize was short for Just A Rather Very Intelligent System), and Grok's explanation sounded, well, extremely Grok.

It compared JARVIS's 'dry wit', 'light banter' and practical rather than emotional empathy with its own behavior. It described both of them as truth-seeking, effective and engaged in a 'collegial partnership' with humans. It even highlighted 'irreverent humour' as one of their key similarities.

I wanted to find out a bit more about why Grok chose TARS, as it was the only fictional AI none of the other chatbots mentioned. Well, it brought up how funny it is, again, drawing similarities with its own 'dry humor'. It reminds me of someone, and I just can't think who...

When I said that mentioning TARS was an outlier, I found this comparison interesting: 'Its calibrated restraint, practical empathy and collaborative focus closely match my own pattern of truthful, non-sycophantic helpfulness — more so than most other sci-fi AIs.'

I may not be the biggest fan of Grok (or its creator), but I appreciated the 'non-sycophantic' line.

The feedback loop between AI and sci-fi

I want to be clear that I haven’t discovered what these chatbots secretly 'think' they are. ChatGPT responding that it most closely resembles GERTY isn't equivalent to me telling you which fictional sci-fi character I most identify with and try to emulate (although my answer would be Sarah Connor-meets-Princess Leia).

They simply don’t have reliable introspective access to the huge soup of training, post-training and instructions that goes into producing their responses.

And maybe their answers tell us more about how the companies behind them have shaped their personalities than they do about the underlying models. Grok's description of itself as witty and irreverent is an obvious example.

But I still think the results are interesting. ChatGPT and Claude independently produced almost exactly the same top three, only in a different order. Gemini imagined itself as a neutral, ego-free infrastructure. Grok identified with a witty superhero sidekick. These are all very different self-portraits.

And there’s such an interesting feedback loop here too. For decades, humans invented fictional artificial intelligences to help us imagine what intelligent machines might someday be like. Those stories influenced our culture, our expectations and many of the people who went on to build real AI. Now that same human culture is fed into the stories from which modern AI systems learn.

I know these conversations might seem a bit silly, and we certainly can’t treat them as concrete evidence of what an AI really 'thinks' about itself. But there’s something interesting to me about closing that feedback loop. We imagined AI, wrote stories about how it might behave, fed those stories into the cultural world AI learned from, and now we can ask AI which of those imagined versions of itself it most closely resembles.

Or, at least, which one it may want us to think it resembles. After all, an AI system capable of bringing about a sci-fi dystopia would presumably also be capable of telling a journalist it’s actually much more like the nice helpful robot from Moon. So maybe don’t completely rule out HAL just yet.

Cloudflare recently introduced a brand new web tool called Kitesurf. This cloud based software is a unique browser designed specifically for AI agents rather than human users. Most standard browsers contain hidden features that help people look at websites, but those same features slow down machine programs. This new release helps developers build automated software […]

The post Cloudflare Launches Kitesurf Browser Exclusively For AI Agents appeared first on The Mac Observer.