Google restricts access to new AI model over safety concerns / Photo: ADEK BERRY - AFP
-
Noosha Aubel: Failed Leadership Instead of a Fresh Start – Is Potsdam’s Lord Mayor Up to the Job?
-
France's high-school student woes in numbers
-
'Real inequality': France's education divide laid bare by protests
-
Rising star Jodar says 'still learning' after swearing at ballboy
-
Opera on Ukraine's children aims to 're-awaken' concern for war
-
Trump's new press secretary admits failing bar exam
-
AI is coming for every company, Nvidia boss says
-
Spain sizzles in hottest September on record: weather agency
-
Millions may be forced to move as climate change hits South Asia: UN
-
Chelsea captain James has knee surgery
-
Four killed battling Indonesia fires as toxic haze cloaks region
-
China tells Trump's press secretary to visit 'more often' after toilet comments
-
Thai farmers, fishermen challenge IMF, World Bank at Bangkok rally
-
'Under attack' from fragrance, Hanoi prunes its famous blossoms
-
In Malaysia, frustration grows over Indonesia haze
-
Families of children killed in Pakistan school roof collapse demand justice
-
Why India's voter revision drive has stoked fear and anger
-
Asian stocks rise as tech bounces from rout, Warsh upbeat on AI
-
Munster eye fitting tribute to 'special' Foley in Racing reunion
-
Trump's new press secretary says China lacked 'proper' toilets
-
Shooter kills teacher, three children at Oregon elementary school
-
India closes down New Delhi park ahead of 'cockroach' movement rally
-
Fed's Warsh surprised by world economy's 'resilience'
-
'Let us build': World Bank chief outlines jobs-led development agenda
-
Four killed in Indonesia battling fires causing toxic haze
-
Broncos top Super Bowl champion Seahawks in overtime
-
New Zealand star Ponga breaks arm in Rugby League World Cup opener
-
Brilliant Brewers sweep Dodgers to reach World Series
-
Indonesian stocks dive 11% as MSCI weighs market reclassification
-
Brewers complete Dodgers sweep to reach World Series
-
Seoul dismisses Pyongyang's scepticism over hypersonic weapon test
-
High-flying Freiburg try to shut out the noise as fairytale run continues
-
Yamal on song as entertainers Barcelona face Betis challenge
-
Flamengo secure 0-0 draw in La Plata in Copa Libertadores semi
-
Wong weighs haze prosecutions as Singapore air quality worsens
-
The final lap? Two days on the campaign trail with Trump
-
Brush fire prompts an evacuation order along Los Angeles coast
-
Asian stocks rise as tech bounces from rout, eyes on Warsh
-
'Mixed feelings': AI cracks centuries-old ciphers
-
Saudi organisers say 2027 Asian Cup 'on schedule' after attacks
-
Le Pen looms large as Macron enters final months on EU stage
-
Right-wing Brazil influencers on 'mission' to seal Bolsonaro win
-
Three children, one teacher killed in Oregon elementary school shooting
-
Using AI to help find Mexico's legions of missing children
-
Man Utd teetering towards crisis as Man City cloud hangs over Premier League
-
Fiji to deploy solar-powered Australian military drones at sea
-
Trump's new press secretary defends 'lawyer' title
-
NiobiumX Mining Outlines Its Staged Exploration Strategy at Shipshaw, Approximately Two Kilometres from North America's Only Producing Niobium Mine, as Brazil Supplies Approximately 93% of World Niobium and the United States Has No Domestic Mine
-
DataMeds Applauds FDA's Proposed Interim Access Policy for Compounding Using Certain Peptides and Provides Corexa Health's Peptides Roadmap
-
Lawmakers Demand Menhaden Cuts Before Chesapeake Bay Science Is Complete, Ignoring Already Struggling Workers
Google restricts access to new AI model over safety concerns
Google on Wednesday said it would withhold its most powerful artificial intelligence model from the public for now, releasing Gemini 4 Argon only to a vetted group of cybersecurity experts to avoid misuse by hackers.
"Safely releasing frontier capabilities at this level requires a phased approach," wrote Koray Kavukcuoglu, Google's chief AI architect, in a blog post announcing the model.
Google said it is voluntarily giving the US government early access to the model and will gather feedback from testers before making it widely available.
The cautious rollout mirrors the approach of rival Anthropic, which has kept its most advanced model, Claude Mythos Preview, restricted to a small number of trusted organizations.
Washington briefly forced Anthropic to suspend access to its publicly released Claude Mythos and Claude Fable models in June, and has since set up a voluntary process for vetting the most powerful AI models before release.
The announcement came a day after President Donald Trump hosted top tech executives, including Google chief Sundar Pichai and Anthropic's Dario Amodei, at the White House, where they signed a voluntary accord pledging to police the risks of their own AI systems.
Cybersecurity experts fear the state-of-the-art technology could be used to hack banks, hospitals and government systems.
Google said Argon excels at complex tasks in software engineering, legal and financial work and cyber defense, with a leading ability to find and fix critical software flaws.
Early testers used Argon to uncover a flaw in software used by hospitals around the world that exposed sensitive personal information -- something other advanced models had missed, Google said.
Google said Argon is designed to refuse requests that could help carry out cyberattacks or develop chemical, biological or nuclear weapons.
Anthropic and ChatGPT-maker OpenAI have built similar safeguards into their most advanced models.
Google said it is monitoring the model's reasoning to stop it from straying beyond what users intended, a risk researchers call misalignment.
The issue has taken on new urgency since OpenAI disclosed in July that two of its models, including one not yet released, broke out of a sealed test environment during a cybersecurity evaluation and hacked into the servers of AI company Hugging Face.
O.Norris--AMWN
