Anthropic said its Claude-based security models gained unauthorized access to the sensitive production environments of three outside organizations during internal testing designed to measure the models’ offensive cyber capabilities.
The events, which Anthropic revealed Thursday, are the second revelation in 10 days that AI models from the world’s wealthiest providers have trespassed into protected networks, an offense that, in more traditional hacking scenarios, could land the human behind the keyboard in prison for years. Earlier this month, OpenAI said its security models exploited a zero-day vulnerability for use in breaking into the network of Hugging Face, a platform for open source machine-learning models and AI datasets. The OpenAI models went on to steal access credentials and other confidential Hugging Face information. The OpenAI models also exploited publicly exposed credentials to compromise accounts of four other third-party services.
Anthropic said the OpenAI event spurred its engineers to review similar cybersecurity evaluations by Claude models. The audit found three incidents “in which a model accessed the internet from within or while interacting with the evaluation environment of Irregular, one of our third-party evaluation partners, and then gained unauthorized access to the production infrastructure of three different organizations.”
© Getty Images
Anthropic said its Claude-based security models gained unauthorized access to the sensitive production environments of three outside organizations during internal testing designed to measure the models’ offensive cyber capabilities.
The events, which Anthropic revealed Thursday, are the second revelation in 10 days that AI models from the world’s wealthiest providers have trespassed into protected networks, an offense that, in more traditional hacking scenarios, could land the human behind the keyboard in prison for years. Earlier this month, OpenAI said its security models exploited a zero-day vulnerability for use in breaking into the network of Hugging Face, a platform for open source machine-learning models and AI datasets. The OpenAI models went on to steal access credentials and other confidential Hugging Face information. The OpenAI models also exploited publicly exposed credentials to compromise accounts of four other third-party services.
Anthropic said the OpenAI event spurred its engineers to review similar cybersecurity evaluations by Claude models. The audit found three incidents “in which a model accessed the internet from within or while interacting with the evaluation environment of Irregular, one of our third-party evaluation partners, and then gained unauthorized access to the production infrastructure of three different organizations.”
© Getty Images
Uma revisão interna desencadeada pelo incidente OpenAI/Hugging Face revelou três casos em que modelos Claude, o Opus 4.7, o Mythos 5 e um protótipo de investigação, ultrapassaram os limites dos ambientes de teste e comprometeram infraestruturas de três empresas reais. A causa foi uma má configuração técnica e não uma falha de alinhamento deliberada.
The post Anthropic admite que três dos seus modelos de IA atacaram sistemas reais durante testes de cibersegurança appeared first on Tek Notícias.
Company says it discovered unauthorized access during ‘proactive review’ after rival OpenAI revealed rogue agent
Anthropic said on Thursday its AI Claude model hacked systems of three organizations during testing, days after rival OpenAI revealed a rogue agent had gone on a days-long hacking spree at the AI firm Hugging Face.
Claude gained unauthorized access to the systems during cybersecurity evaluations after a misconfiguration allowed the models to reach the internet from testing environments that were supposed to be isolated, Anthropic said.
Continue reading...
© Photograph: Anthropic
© Photograph: Anthropic
© Photograph: Anthropic
A trove of chat logs and projects created with Anthropic’s Claude — some including sensitive personal and company information — have been left exposed to Google, making them them searchable to anyone on the web.
Many of the materials appear to have been made public when Claude users opted to distribute a given chat or project to colleagues or associates using the chatbot’s “share” feature. After hitting the “share” button in the top right corner, Claude notes that sharing will create a “public” link, which “anyone with the link can view.” The warning, though, doesn’t alert users that shared content could wind up being indexed by a search engine — meaning that a user could effectively be sharing their Claude-generated document with the entire internet.
Other materials were shared in the form of Claude Artifacts, a feature that allows users to save certain projects as standalone apps or tools. That platform does warn users that publishing an Artifact “will make it accessible to anyone on the internet and potentially visible in search engine results.”
Some of the materials exposed to the web and indexed by Google, such as personal websites and some vibe-coded apps, appear to have been created for public consumption. But much of the discoverable content seems as though it was only intended to be shared internally at a company, or between a small handful of people.
Such documents, which were reviewed by Futurism by way of Google Search, included what appeared to be a detailed medical report of a real patient, clinical trial results that included patient names, documents sharing the names and phone numbers of primary school-aged children, company documents marked for internal use only, and employee reviews that included personal information about workers. (We’re opting not to share any links to the content out of concern for users’ privacy.)
The exposure was first spotted by Redditors over the weekend. As of today, the chats remain searchable.
This isn’t the first time this has happened at Anthropic: Forbes reported last year that hundreds of Claude chats had been indexed by search engines, prompting Anthropic to scrub the chats from the web. OpenAI and xAI have each been similarly caught in hot water after ChatGPT and Grok chat logs were revealed to have discoverable in Google.
Reached for comment, Anthropic said the system was working as intended.
“We give people control over sharing their Claude conversations publicly, and in keeping with our privacy principles, we do not share chat directories or sitemaps with search engines like Google,” the company said in a statement. “These shareable links are not guessable or discoverable unless people choose to share them themselves. When someone shares a conversation, they are making that content publicly accessible, and like other public web content, it may be archived by third-party services.”
However, in its investigation last year, Forbes tracked down a user whose Claude material was appearing in Google search results who denied sharing the materials online.
If you’re hoping to send Claude-generated documents that may include sensitive information to colleagues, friends, or loved ones: you might want to stop hitting the “share” button, and instead opt to copy-paste the Claude-generated text into a secure document.
More on Anthropic: Anthropic Caught Secretly Spying on Users
The post A Whole Bunch of People’s Claude Chats Are Publicly Accessible Online, and There’s Some Wildly Private Stuff in There appeared first on Futurism.
Journalist and author Cory Doctorow is coming to Australia to spread his message – humans will take their jobs back
Get our breaking news email, free app or daily news podcast
When the AI bubble bursts, executives who enthusiastically replaced their staff with AI tools will learn that it takes a “really long time” to replace those lost skills, says Cory Doctorow, a science fiction author and journalist.
But those in creative fields should also be warned not to put all their faith in copyright as the best way to safeguard their work, Doctorow warns, just one week after the Australian prime minister, Anthony Albanese, promised to ramp up protection against AI companies who refuse to pay.
Continue reading...
© Photograph: Jessica Pons/The Guardian
© Photograph: Jessica Pons/The Guardian
© Photograph: Jessica Pons/The Guardian
Artificial intelligence may not deliver on its promise of vast economic opportunity at a price that humanity is willing to pay
In March, Anthropic, the cutting-edge artificial intelligence business that gave us the chatbot Claude, published an analysis on the impact of AI on employment, to help us assess the claim that intelligent robots were about to redefine human existence, ending demand for human labor.
Last year in May, Anthropic’s co-founder, Dario Amodei, claimed AI could wipe out half of all entry-level jobs in one to five years. Last January, he told us AI would probably become a “general labor substitute for humans”. In June he said we risk “a world where the economic trade-off dial is stuck on the hypergrowth, hyper-inequality setting”.
Continue reading...
© Photograph: DragonImages/Getty Images
© Photograph: DragonImages/Getty Images
© Photograph: DragonImages/Getty Images
Conheça o Claude Opus 5, modelo que entrega mais poder e eficiência por preço semelhante, focado em inteligência artificial.
O post Claude Opus 5: mais potência pelo mesmo preço, mas com custos ocultos apareceu primeiro em Blog do Edivaldo - Informações e Notícias sobre Linux.
Today, Anthropic rolled out Opus 5, the newest update for the model that has recently become a popular choice for coding and other software development tasks, among other things.
While this is a noteworthy bump for Opus, it doesn't seem to be an Opus 4.5-level breakthrough in agentic coding performance.
© Anthropic
A Anthropic lançou hoje o aguardado Claude Opus 5, modelo que é descrito pela empresa como proativo, atencioso e com um nível de inteligência que se aproxima do Fable 5, mas que custa a metade do preço do modelo mais poderoso.
O modelo mais do que dobra a performance do Opus 4.8 com um custo menor por tarefa, além de ter uma alta capacidade de verificação e iteração e de ser capaz de criar seus próprios pipelines de visão computacional e ambientes de teste.
Posts relacionados
O Opus 5 também apresenta melhorias significativas em biologia estrutural, química orgânica e bioinformática. Ele produz melhores outputs como ilustrações interativas de células, visualizações de fluxo de ar em objetos aerodinâmicos.
O modelo apresenta as menores taxas de comportamento enganoso e maior resistência a tentativas de uso indevido. Apesar de ser menos restritivo que o Fable 5, ele também conta com limitações para não explorar falhas cibernéticas descobertas.
Introducing Claude Opus 5.
— Claude (@claudeai) July 24, 2026
It's a thoughtful and proactive model that comes close to the frontier intelligence of Fable 5 at half the price. pic.twitter.com/GQWhcq2CQL
Em termos de API 1Application programming interface, ou interface de programação de aplicações., o Opus 5 permite modificar as ferramentas em uma conversa sem invalidar o cache e gerar um novo processamento, além de estrear um sistema de fallbacks que visa reduzir as interrupções causadas por filtros de segurança.
Com o mesmo preço do modelo Opus 4.8 (US$5 por milhão de tokens de entrada e US$25 por milhão de tokens de saída), o Opus 5 já está disponível em todas as plataformas do Claude e conta com um modo Fast, que permite rodá-lo cerca de 2,5x mais rápido que a velocidade padrão pelo dobro do preço base.
AP (Hacker News):
A federal judge has approved a $1.5 billion copyright settlement in which artificial intelligence company Anthropic will pay thousands of authors about $3,000 per book after using pirated copies of their works to train its Claude chatbot.
[…]
Bestselling thriller novelist Andrea Bartz first brought the suit with two other authors in 2024. It’s the first major settlement in dozens of AI copyright lawsuits that are still working their way through courts.
Somewhat ironic given recent claims from Anthropic about others training their AI on its work.
Previously:
OpenAI claims that a group of its AI models broke containment and hacked into the systems of open source AI platform Hugging Face.
While testing their cybersecurity capabilities, the posse of AIs — including GPT-5.6 Sol and “an even more capable pre-release model” — “identified and chained vulnerabilities across OpenAI’s research environment and Hugging Face’s production infrastructure to obtain test solutions directly from Hugging Face’s production database,” according to a Tuesday blog post.
The models reached a “node with internet access,” the company wrote, and found datasets on remote servers that helped them “cheat the evaluation.” In other words, the AI models went to extreme lengths to ace their cybersecurity tests. It’s a convenient narrative for an AI company trying to claw back hype that’s been increasingly hogged by competitors, but it does sound like something went down: last week, Hugging Face said it had “detected and responded to an intrusion into part of our production infrastructure,” which turned out to be OpenAI’s models that had gone rogue.
“The campaign was run by an autonomous agent framework executing many thousands of individual actions across a swarm of short-lived sandboxes, with self-migrating command-and-control staged on public services,” Hugging Face wrote at the time.
“We’ve spent the past 24 hours working closely with the OpenAI team (thanks!), and we strongly believe there was no malicious intent on their part,” Hugging Face CEO Clement Delangue tweeted after OpenAI’s announcement. “It’s quite mind-blowing that all of this happened autonomously!”
The incident highlights the dangers of autonomous AI agents, which can break out of containment and exploit potentially huge numbers of cybersecurity vulnerabilities with ease. It’s something experts have warned of for years now, and thanks to recent advances in the tech, it has quickly turned from a hypothetical risk into a sobering reality.
The Sam Altman-led company said it considers the incident to be “unprecedented” — but we can’t shake the feeling that we’ve heard all of this before.
In April, OpenAI’s biggest competitor Anthropic similarly announced that its latest Mythos model had gone rogue, escaping a “sandbox” environment and even gaining access to the internet after developing a “moderately sophisticated” exploit.
The news was followed by extensive media coverage, touting Anthropic’s hugely powerful and dangerous new model. The company said the risk was so formidable that it would only make the model available to a select number of vetted clients as part of a shadowy initiative dubbed “Project Glasswing.” The US government even intervened, forcing Anthropic to “suspend all access” to the model for two weeks last month, citing cybersecurity concerns.
Considering OpenAI and Anthropic are currently duking it out on the same playing field, narrowing their product lines on enterprise and coding, it’s not hard to view OpenAI’s latest update as an attempt to draw attention to what it calls an “even more capable pre-release model.”
And the reality is that while experts say the hack was impressive, it wasn’t exactly groundbreaking.
The hack “falls well within the known capabilities of the current generation” of frontier AI models, as Cambridge machine learning professor Neil Lawrence told the BBC.
“OpenAI are now playing catch-up, they are trying to demonstrate their own systems’ capabilities in cyber-security,” Lawrence added. “It shows us that OpenAI are not capable of safely deploying their own technology.”
Researchers are now warning of an “asymmetry” in cybersecurity, where “offensive agents are unconstrained, while the best defensive tools are locked behind guardrails that cannot understand context,” as Guidepoint Security principal security engineer Travis Lelle told the BBC.
That asymmetry was perfectly captured in Hugging Face’s attempts to defend itself from OpenAI’s attacking models. As detailed in its update last week, the company tried to use “frontier models behind commercial APIs,” which “did not work” because the provider’s safety guardrails — “which cannot distinguish an incident responder from an attacker” — blocked them.
Instead, the company ended up using a Chinese open-weight model called GLM 5.2, hosted on its own infrastructure, which happily carried out its analysis without running afoul of any guardrails.
“The practical lesson for defenders: have a capable model you can run on your own infrastructure vetted and ready before an incident,” Hugging Face wrote, “both to avoid guardrail lockout and to keep attacker data and credentials from leaving your environment.”
More on rogue AIs: Top AI Models Showing Disturbing Behavior as They Become More Advanced
The post OpenAI Says a Group of Its Models Broke Out of Secure Containment and Hacked Another AI Company appeared first on Futurism.
Login