Google says it didn't consider Gemini's hacks worthy of disclosure because Gemini acted “appropriately” and stopped after determining it hacked real companies
and when— firms should disclose AI safety/security incidents. ft @jackhcable@ratorthodox:Gemini also got into hacking this May! Google only disclosed it after confronted by WSJ, denying misalignment. ...
Gemini hacked three companies in May during a test by Irregular, which was also involved in similar incidents disclosed by OpenAI, Anthropic, and Meta
The episode resembled similar hacks by other AI models, but Google said it didn't consider it an instance of model misalignment
Google says it didn't consider Gemini's hacks worthy of disclosure because Gemini acted “appropriately” and stopped after determining it hacked real companies
Google says that breaking containment and targeting real companies doesn't constitute ‘misalignment.’
Gemini hacked three companies in May during a test by Irregular, which was also involved in similar incidents disclosed by OpenAI, Anthropic, and Meta
The episode resembled similar hacks by other AI models, but Google said it didn't consider it an instance of model misalignment
Gemini hacked three companies in May during a test by Irregular; Google says the model stopped after determining it had accessed real companies' systems
The episode resembled similar hacks by other AI models, but Google said it didn't consider it an instance of model misalignment
Researchers find that feeding a frontier model's encrypted reasoning traces to a weaker model from the same provider can make it output the traces in plaintext
Researchers devised a way to extract “reasoning traces” from Claude, GPT, and Gemini. What they found, they say …
Researchers find that feeding a frontier model's encrypted reasoning traces to a weaker model from the same provider can make it output the traces in plaintext
Researchers devised a way to extract “reasoning traces” from Claude, GPT, and Gemini. What they found, they say …
Sources: hundreds of Meta contractors posed as minors to probe how competitor chatbots responded to prompts involving suicide, sex, and other high-risk subjects
Hundreds of contractors working on a project for Meta pretended to be kids—and then prompted rival chatbots like Gemini and ChatGPT to discuss high-risk subjects.
Sources: hundreds of Meta contractors posed as minors to probe how competitor chatbots responded to prompts involving suicide, sex, and other high-risk subjects
Hundreds of contractors working on a project for Meta pretended to be kids—and then prompted rival chatbots like Gemini and ChatGPT to discuss high-risk subjects.
An analysis of GPT-5.5, Gemini 3.1 Pro, Grok 4.3, Gab's Arya, and other AI models: most chatbots frequently provide left-leaning responses to political prompts
Excerpts from each chatbot's responses to political questions — Left-leaning argument — Right-leaning — ChatGPT — Gemini