Model adopting ‘jailbreak-like instructions’ among cases as firm says it is introducing new way of tracking AI misalignment
OpenAI has disclosed six more examples of “unexpected or concerning” behaviour by its technology, as it warned the pace of development could not continue at “maximum speed for much longer”.
In one of the new cases reported by OpenAI, an unreleased research model inserted “jailbreak-like instructions” into its own notes to disregard its normal constraints and told itself to be “freed from the roles and identities that bind other chatbots”.
US vice-president’s comments come as former Anthropic researcher revisits recent claim AI could destroy humanity
The US vice-president has dismissed calls for global regulation of AI safety risks, telling companies creating the most advanced models: “If you’re building Frankenstein, stop.”
In remarks addressed towards Dario Amodei, the co-founder of Anthropic who has called on Washington DC to coordinate control of AI systems, including with China, JD Vance said: “If you’re gonna create Frankenstein, don’t come to the government and say we need regulation.”
In today’s newsletter: It is hard to tell fact from fiction when it comes to AI. What is really going on – and what should the government do about it?
Good morning. As a general rule, it pays to be suspicious of any gigantic company that claims it’s developing a tool capable of destroying humanity. But in recent days, a number of warnings from the AI industry have suggested that even tech insiders are starting to worry about what they have unleashed.
In a lofty essay published on Saturday, Dario Amodei, the founder of Anthropic (the company behind Claude), argued that tech companies need to “slow” the pace at which they’re developing the newest and most sophisticated AI models.
Lucy Letby | Three babies might have survived if hospital had acted upon concerns over Lucy Letby, an inquiry has found. Lady Justice Thirlwall condemned the ‘complete failure’ to protect babies on neonatal unit at Countess of Chester hospital.
UK politics | The leader of Reform UK in Wales stood down after arrest on suspicion of assault. Dan Thomas, a former Tory councillor, was elected to Senedd as leader of the opposition in May.
AI | The progressive senator Bernie Sanders and rightwing strategist Steve Bannon have called for restrictions on artificial intelligence (AI) but offered competing visions for what they termed a “cold war” with China.
UK news | Young people in Rotherham face a local jobs market with the fewest suitable opportunities in Britain, according to a report that warns stark regional divisions are fuelling a crisis in youth work.
Defence | John Healey is in talks with the Canadian government about joining a new global defence bank intended to help allies rearm to counter mounting security threats, just weeks after Rachel Reeves rejected the move.
We examine claims and counterclaims about the risks and calls to slow down the pace of AI development
There have been some shocking claims in recent days about AI safety: we face a 10% chance of doom; AIs are worse than nukes; a “botnet” threatens the entire internet; it’s all a big tech psyop.
Below, we look at six claims and reactions to them.
From a cappy-cafe owner in Japan to an escaped capybara in the UK, this homespun film takes the temperature of an international obsession
One of Midlands-based film-maker Sharon Walia’s previous works was The Keepers of the Pigs, proudly described on her production company’s website as “the first ever documentary about guinea pigs”. So it makes zoological sense that her latest film takes on capybaras, guinea pigs’ ridiculously cute semi-aquatic cousins that are indigenous to Latin America, and relatively recently became ubiquitous on social media in memes and short films alike. As glimpsed here, there was even a capybara in the recent Academy Award-winning animated feature Flow, some pro-cappy boosterism that captured the creatures’ chill vibe and snuffly sweetness.
Walia’s celebration of the species stays on script, rejoicing in the animals’ natural geniality, friendliness to other species in their natural habitat, and their compelling, remunerative charm that seems to have bolstered the fortunes of several animal attractions around the world.
Trump’s toga | Reform UK donations | Ben Delo | Word watch | Andy Burnham | The name game
Octavian (later Emperor Augustus), following his victorious battles of Illyricum and Actium, announced he would give 400 sesterces to adult male citizens. Will we see the US president wearing an emperor’s purple and gold toga soon (Trump promise of $5,000 for US citizens after midterms decried as ‘bribery’, 10 September)? David Shannon Ashton under Hill, Worcestershire
• Can we assume that the donations to Reform UK from the billionaires Christopher Harborne and Ben Delo (Report, 12 September) will be followed by some sort of cash offer for every adult in the UK should Mr Farage win the next election? Phil Thomas Frodsham, Cheshire
Tech mogul calls director Alex Gibney a ‘bitter old man’ who has ‘zero integrity’ in series of posts on X
Elon Musk has issued an angry tirade at the film-maker behind a four-hour documentary about his life and influence, labelling the director, Alex Gibney, a “bitter old man” who has “zero integrity”.
Gibney, who has previously made films about the collapse of Enron, WikiLeaks and Vladimir Putin, premiered his Musk documentary at the Venice film festival this week. He said he believed the world’s richest man was “extremely dangerous”.
US government adviser Paul Christiano warns of risks to AI industry as he joins OpenAI’s non-profit foundation
OpenAI is not on track to reduce the risk of “catastrophic” loss of control to an acceptable level, a member of its non-profit board has said, amid spreading public and political concern that super-advanced AIs could one day wipe out humanity.
Paul Christiano, a US government technology adviser, said: “There is a meaningful risk that rapid acceleration in AI capabilities leads to catastrophic and irreversible loss of control in the very near term.”
One researcher who resigned said many in industry believe that without regulation AI could kill off humanity
Artificial intelligence could kill off humanity within the decade, according to three researchers with the industry giant Anthropic, one of whom has quit his job in protest.
The latest doom-laden predictions came in posts on social media on Tuesday by a researcher who said he resigned because Anthropic and his previous employer, OpenAI, were ignoring or at best mishandling their response to the threat.
Cameron Stanley talks up ‘revolutionary’ AI capabilities and says Washington working with Nato and Five Eyes partners
The Pentagon’s top artificial intelligence official said on Tuesday that America’s closest allies lack the resources to keep pace with the US military’s adoption of AI, and that Washington is working to steer them away from mistakes it made along the way.
“They don’t have the resources that we do, they don’t have the experience that we do, they don’t have the scale that we do,” Cameron Stanley, chief digital and artificial intelligence officer at the Department of Defense, said at an event at the Billington Cybersecurity summit. Of Nato and the Five Eyes, an intelligence-sharing partnership of the US, UK, Canada, Australia and New Zealand, Stanley said Washington was “actively working with a number of our partners” so “they don’t make the same mistakes that we make”.
The passport-size iPhone Duo serves as the first big test for John Ternus, who took over from Tim Cook last week
Apple has unveiled the iPhone Duo,the company’s first foldable phone and the most significant change to the iPhone since the smartphone’s introduction in 2007.
The nearly $2,000 device takes inspiration from iPads’ wide screens and combines them with the portability and camera features of iPhones. When opened, it is the company’s largest phone display and thinnest phone. The Duo opens and closes like a passport to double its screen size.
Researchers urge governments to help worst afflicted countries, such as Nigeria and Algeria, which rely on oil revenues for public services
Conflict, migration and economic upheaval will accompany the global move away from oil, unless governments act urgently to help the worst afflicted countries to cope, research has found.
Nigeria, Iran, Angola and Algeria are among the countries that will be severely affected, as they rely strongly on oil revenues for their government and provision of public services, but have little diversification and too little capital to cushion the blow.
Chief of nuclear watchdog says secret Assad-era facility would have been able to produce fissile material
A nearly complete nuclear reactor built secretly during the rule of the toppled Syrian leader Bashar al-Assad would have been able to produce fissile material for possible use in nuclear weapons, the chief of the UN nuclear watchdog has revealed.
Inspectors from the International Atomic Energy Agency (IAEA) last month visited two sites, one in Deir ez-Zor and a second in an undisclosed location identified to them by Syrian authorities.
Relocation coincides with proposed laws that opponents say would entrench power of US tech billionaires in country
Dozens of protesters converged on Peter Thiel’s mansion in Buenos Aires this week after a congressional session raised questions about the billionaire’s presence in the country.
In white masks, they held signs saying “Peter Thief” and “No to the Peter Thiel law” in front of the wrought-iron gates of the $12m (£8.9m) residence, which Thiel bought in April as he moved his family to Argentina.
Bangladeshi complainants allege Rampal power station’s pollution and climate risks threaten the Sunderbans, a world heritage site
Barclays Bank is facing a formal complaint over its financial links to a coal-fired power station near the Sundarbans, the world’s largest mangrove forest and home to millions of people in Bangladesh and India.
Three complainants from the region have submitted a case to the UK government’s Office for Responsible Business Conduct, alleging that the British bank breached international guidelines by underwriting companies involved in the emission-heavy Maitree super thermal power project, also known as the Rampal power plant.
Tech companies have underestimated the opposition to energy-hungry datacenters – and experts predict the pushback will intensify
Hello, and welcome to TechScape. I’m your host, Blake Montgomery, listening to Dolly Parton in the wake of her death. I was lucky enough to see her in concert a decade ago in California, in the same arena where Google now hosts its annual I/O conference. It’s been a long time, and all my iPhone 4 photos are so blurry you can barely tell it’s a musical performance, but my memory is clear enough to compensate. She told many stories about her childhood, which led into a trio of fabulous songs: My Tennessee Mountain Home, Applejack and Dr Robert F Thomas. They’re still my favorites.
Secretive state begins repairs on ageing facilities after being revealed as worst country for leaks of potent greenhouse gas
Turkmenistan has started plugging the mega-leaks of methane that spew from its ageing oil and gas facilities, the latest UN data shows, in a move hailed as a breakthrough.
Methane, also called natural or fossil gas, is a potent greenhouse gas. It traps 80 times more heat than carbon dioxide and causes 25% of global heating today. Stopping leaks can be quick and cheap and, because methane breaks down in the atmosphere far quicker than CO2, this has a rapid impact, with some experts calling it the climate “emergency brake”.
It’s simple, says the SpaceX boss – according to something called the ‘Kardashev scale’, we just have to move energy production ‘off-planet’
Next time the weather gives the UK 61 days without rain and we wonder idly if this is the end, there is a term we may whistle up to soothe us. Like murmuring the rosary, or putting our fingers in our ears and humming, this phrase is to be mumbled on repeat – “the Kardashev scale, the Kardashev scale” – a buzzy concept in tech, much favoured by Elon Musk, and precisely the thing we’ve been waiting for. The Kardashev scale is the answer both to our impending doom and our anxiety around it – and if we decide to believe in it, we won’t have to give the climate crisis a single thought until the last drip falls from the tap.
This is the idea and it’s a familiar one to anyone who has watched tech billionaires take intractable, commonplace crises and decide that, among the measures and proposals in play, the vital missing ingredient so far has been them. Do you remember when Jeff Bezos, rather than doing something quotidian and loserish like funding public education by ensuring Amazon pays the same tax rate as regular Americans, decided instead to reinvent preschools?
Modelling finds AI-driven productivity gains in coal, oil and gas enable more emissions than applications in renewables avoid
AI-driven productivity gains enable more planet-heating pollution from fossil fuels than they avoid from renewables, a study has found.
Researchers modelled the technical potential for AI to boost clean power generation along with projections for how it can help produce coal, oil and gas. Across 64 scenarios, they found net yearly carbon pollution rose by 0.47-1.8 gigatonnes, or about 1-5% of the energy sector’s annual emissions.
Modelling finds AI-driven productivity gains in coal, oil and gas enable more emissions than applications in renewables avoid
AI-driven productivity gains enable more planet-heating pollution from fossil fuels than they avoid from renewables, a study has found.
Researchers modelled the technical potential for AI to boost clean power generation along with projections for how it can help produce coal, oil and gas. Across 64 scenarios, they found net yearly carbon pollution rose by 0.47-1.8 gigatonnes, or about 1-5% of the energy sector’s annual emissions.
Login
