AI pioneer announces non-profit to develop ‘honest’ artificial intelligence

A picture


An artificial intelligence pioneer has launched a non-profit dedicated to developing an “honest” AI that will spot rogue systems attempting to deceive humans,Yoshua Bengio, a renowned computer scientist described as one of the “godfathers” of AI, will be president of LawZero, an organisation committed to the safe design of the cutting-edge technology that has sparked a $1tn (£740bn) arms race,Starting with funding of approximately $30m and more than a dozen researchers, Bengio is developing a system called Scientist AI that will act as a guardrail against AI agents – which carry out tasks without human intervention – showing deceptive or self-preserving behaviour, such as trying to avoid being turned off,Describing the current suite of AI agents as “actors” seeking to imitate humans and please users, he said the Scientist AI system would be more like a “psychologist” that can understand and predict bad behaviour,“We want to build AIs that will be honest and not deceptive,” Bengio said.

He added: “It is theoretically possible to imagine machines that have no self, no goal for themselves, that are just pure knowledge machines – like a scientist who knows a lot of stuff.”However, unlike current generative AI tools, Bengio’s system will not give definitive answers and will instead give probabilities for whether an answer is correct.“It has a sense of humility that it isn’t sure about the answer,” he said.Deployed alongside an AI agent, Bengio’s model would flag potentially harmful behaviour by an autonomous system – having gauged the probability of its actions causing harm.Scientist AI will “predict the probability that an agent’s actions will lead to harm” and, if that probability is above a certain threshold, that agent’s proposed action will then be blocked.

LawZero’s initial backers include AI safety body the Future of Life Institute, Jaan Tallinn, a founding engineer of Skype, and Schmidt Sciences, a research body founded by former Google chief executive Eric Schmidt.Sign up to Business TodayGet set for the working day – we'll point you to all the business news and analysis you need every morningafter newsletter promotionBengio said the first step for LawZero would be demonstrating that the methodology behind the concept works – and then persuading companies or governments to support larger, more powerful versions.Open-source AI models, which are freely available to deploy and adapt, would be the starting point for training LawZero’s systems, Bengio added.“The point is to demonstrate the methodology so that then we can convince either donors or governments or AI labs to put the resources that are needed to train this at the same scale as the current frontier AIs.It is really important that the guardrail AI be at least as smart as the AI agent that it is trying to monitor and control,” he said.

Bengio, a professor at the University of Montreal, earned the “godfather” moniker after sharing the 2018 Turing award – seen as the equivalent of a Nobel prize for computing – with Geoffrey Hinton, himself a subsequent Nobel winner, and Yann LeCun, the chief AI scientist at Mark Zuckerberg’s Meta.A leading voice on AI safety, he chaired the recent International AI Safety report, which warned that autonomous agents could cause “severe” disruption if they become “capable of completing longer sequences of tasks without human supervision”.Bengio said he was concerned by Anthropic’s recent admission that its latest system could attempt to blackmail engineers attempting to shut it down.He also pointed to research showing that AI models are capable of hiding their true capabilities and objectives.These examples showed the world is heading towards “more and more dangerous territory” with AIs that are able to reason better, said Bengio.

trendingSee all
A picture

Sports Direct pricing practices ‘may be breaking the law’, Which? says

Sports Direct could be breaking the law by misleading shoppers into thinking they are getting a good deal, a consumer body has claimed, after it looked at prices of items ranging from trainers to hoodies.Which? said it had reported the retailer to the Competition and Markets Authority after uncovering what it claimed were “some questionable and dodgy pricing tactics” on its website.The organisation said it had found products being sold on SportsDirect.com with recommended retail prices (RRPs) “that appear to be misleading”, as its researchers could not find the products sold at that RRP price anywhere else online.It meant people may be being misled “into thinking they are getting a better deal than they really are”

A picture

Bonuses banned for 10 English water bosses over sewage pollution

Bonuses for 10 water company executives in England, including the boss of Thames Water, will be banned with immediate effect over serious sewage pollution, as part of new powers brought in by the Labour government.The top executives of six water companies who have overseen the most serious pollution events will not receive performance rewards this year, the environment said.The companies – Thames Water, Anglian Water, Southern Water, United Utilities, Wessex Water and Yorkshire Water – are responsible for the most serious category of sewage pollution into rivers and seas, all of which are, or have been, under criminal investigation by the Environment Agency.Under powers in Labour’s Water (Special Measures) Act 2025, the regulator, Ofwat, is now able to ban bonuses for water executives where a company fails to meet key standards on environmental and financial performance, or is convicted of a criminal offence.In the past 10 years, executives at the nine main water and sewerage companies have been paid £112m in bonuses while sewage pollution increased to a record last year of 2,487 events

A picture

Tesla share plunge amid Trump feud wipes $152bn off Elon Musk’s company

Tesla’s shares dropped by about 14.2% on Thursday at market close, wiping roughly $152bn off the value of the company as a feud between Elon Musk and Donald Trump erupted into public view. The former political allies traded threats and insults through posts on their respective social media platforms throughout the afternoon as the company’s price fell.Trump suggested on Truth Social that he could cut Musk’s government subsidies and contracts, of which both Tesla and SpaceX have been immense beneficiaries. Musk meanwhile threatened to decommission the SpaceX spacecraft that Nasa relies on for transport missions, called for Trump’s impeachment, derided the president’s signature tariffs and accused him of being affiliated with the notorious sex offender Jeffrey Epstein

A picture

23andMe back on the auction block after former CEO makes 11th-hour bid

The DNA testing company 23andMe is back up for sale, throwing a purchase agreement reached last month into chaos, court filings show.The board of directors of 23andMe, which filed for bankruptcy in March, had agreed to sell the company and its assets to the pharmaceutical firm Regeneron for $256m after conducting an auction in April. However, the founder and former CEO of the genetics company, Anne Wojcicki, put in a $305m bid through a newly formed non-profit, TTAM Research Institute, after the auction ended and pushed the bankruptcy court to reopen the sale process. She tried to buy the company multiple times during its long decline and bankruptcy but was rejected by the board.TTAM’s offer of $305m will serve as a starting price for the secondary sale process, and Regeneron will be permitted to submit a competing bid that is at least $10m more

A picture

‘A great privilege’: Mal Meninga locked in as Perth Bears’ inaugural NRL coach

The Perth Bears hope the presence of Mal Meninga will give the NRL’s 18th team immediate cut-through in an AFL-dominated city after unveiling the Immortal as the head coach of the start-up franchise.At a press conference in Sydney on Friday, Meninga was locked in as the Bears’ inaugural coach on a three-year deal. It is his first foray into club coaching in more than 25 years.Meninga has renounced his role as coach of the Australian Test team ahead of an end-of-season Ashes tour. The 64-year-old will now set about building a competitive roster for the Bears’ first NRL season in 2027

A picture

Aaron Rodgers ends time in wilderness by signing with Pittsburgh Steelers

The NFL’s most nagging storyline has ended with the news that Aaron Rodgers is has agreed to sign a one-year contract with the Pittsburgh Steelers.The 41-year-old Rodgers parted ways with the New York Jets earlier this year, and had been linked for some time with the Steelers, who were without an established starter at quarterback. But Rodgers had given mixed signals about whether he wants to continue his career.“I’m in a different phase of my life,” Rodgers said in April. “To make a commitment to a team is a big thing