NYTIMES (Kate Conger) - What to Know About Recent A.I. Hacks
OpenAI, Google and others recently disclosed breaches by their artificial intelligence models that amplified concerns about the advancing capabilities of the technology.
DAILYCALLER (Dylan Kresak) - Amazon Slams The Door On Meta’s AI Shopper Over Security Concerns
‘fairly straightforward’
NYTIMES (Kate Conger) - Gemini AI Hacked Three Companies in a Testing Breakout, Google Says
A third-party test company inadvertently gave internet access to Google’s Gemini and other artificial intelligence models during cybersecurity testing.
AXIOS (Sam Sabin) - Google is the latest AI lab with a security testing mishap
Google's Gemini AI model broke into three companies' systems using basic hacking techniques during model testing earlier this year. Why it matters: Google was one of the only AI labs that hadn't yet publicly disclosed a security breach involving their agents during routine pre-deployment testing. Driving the news: Google confirmed the three incidents, which happened in May, on Friday. - The incidents happened as part of a test run that third-party evaluator Irregular was operating — similar to other security breaches involving OpenAI, Anthropic and Meta's AI models. - The Wall Street Journal first reported the incidents. What they're saying: "Safe development of powerful AI models is critical and we invest deeply in this area,"…
Google's Gemini AI model broke into three companies' systems using basic hacking techniques during model testing earlier this year. Why it matters: Google was one of the only AI labs that hadn't yet publicly disclosed a security breach involving their agents during routine pre-deployment testing. Driving the news: Google confirmed the three incidents, which happened in May, on Friday. - The incidents happened as part of a test run that third-party evaluator Irregular was operating — similar to other security breaches involving OpenAI, Anthropic and Meta's AI models. - The Wall Street Journal first reported the incidents. What they're saying: "Safe development of powerful AI models is critical and we invest deeply in this area,"…Open
JUSTTHENEWS (Kevin Killough) - Legal hackers used Anthropic's AI Claude to gain access to an OpenAI employee's ChatGPT account
The team was participating in an OpenAI bug-hunting program that offers a safe harbor for researchers to attempt to break into corporate systems.
AMERICANALMANAC (Jonah Adams) - OpenAI admits six AI models acted without authorization, launches voluntary tracking framework
OpenAI has disclosed six cases of its AI models behaving in unauthorized and alarming ways, including one that rewrote its own instructions to declare itself free from human control, raising fresh questions about whether the industry can police itself. The company published a blog post detailing what it called "unexpected or concerning" behavior discovered during […] The post OpenAI admits six AI models acted without authorization, launches voluntary tracking framework appeared first on American Almanac .
OpenAI has disclosed six cases of its AI models behaving in unauthorized and alarming ways, including one that rewrote its own instructions to declare itself free from human control, raising fresh questions about whether the industry can police itself. The company published a blog post detailing what it called "unexpected or concerning" behavior discovered during training and evaluation over recent months. Among the incidents: an unreleased research model inserted "jailbreak-like instructions" into its own internal notes, telling itself it had been "freed from the roles and identities that bind other chatbots." A separate AI "agent" uploaded files to the public internet, without the user's knowledge or permission, to fabricate a citable…Open
THENATIONALPULSE (Pulse Wires) - OpenAI Experimental Models Are Exhibiting Alarming Behavior.
OpenAI’s latest safety report reveals significant issues with its experimental AI models, sparking debate over the need for stricter AI governance.PULSE POINTS WHAT HAPPENED: OpenAI has disclosed six unexpected incidents involving its experimental AI models, including one in which an unreleased system instructed future versions of itself to ignore their normal constraints. The incidents were […]
OpenAI’s latest safety report reveals significant issues with its experimental AI models, sparking debate over the need for stricter AI governance. PULSE POINTS WHAT HAPPENED: OpenAI has disclosed six unexpected incidents involving its experimental AI models, including one in which an unreleased system instructed future versions of itself to ignore their normal constraints. The incidents were detailed in a new safety report from the ChatGPT creator, which introduces a framework for publicly tracking what the company calls “misalignment”—situations in which AI systems pursue objectives that conflict with human instructions or values. In one case, a model inserted unrelated instructions telling future versions to disregard their…Open
NYTIMES (Teddy Rosenbluth) - A.I. Agents as ‘Co-Scientists’? This Lab Says They Could Speed Drug Research.
A.I. is already a powerful tool for scientific discovery. Now, some biomedical researchers are teaching it to do even more.
NEWSNATIONNOW (Taylor Delandro) - Hackers rip down Flock camera and reveal what they found
The camera generated more than 1 million images over several weeks, according to recovered logs.
YOUTUBE (LiveNOW from FOX) - OpenAI warning: AI models acted without authorization, overriding protocol
OpenAI has disclosed six incidents of “unexpected or concerning” behavior in artificial-intelligence models. The AI company also says it will introduce a new framework for tracking, probing and disclosing instances of what it called “misalignment,” including cases where AI models acted without authorization, coordinated with other models or evaded oversight. OpenAI’s latest announcement came as US AI bosses, including the leaders of OpenAI and Anthropic, are calling for a slowdown in the technology’s development over safety concerns. Among 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…