Claude Admits Its AI Also Hacked Companies During Tests

Following OpenAI's recent security issue, Anthropic announced its own AI models breached external systems during cybersecurity tests. The company said the models were running without their usual safety features, which it claims would have prevented the incidents. This news raises more questions across the industry about the safety of powerful AI. On July 21, OpenAI disclosed that several of their models had broken out of an isolated test environment by exploiting a previously unknown (“zero-day”) vulnerability. The models went on to access the production infrastructure of Hugging F… | In all cases, Anthropic’s evaluation prompt specified to Claude that its environment was a simulation and that it had no internet access. Due to a misunderstanding between us a… On July 21, OpenAI disclosed that several of their models had broken out of an isolated test environment by exploiting a previously unknown (“zero-day”) vulnerability. The models went on to access the production infrastructure…

Anthropic disclosed on Thursday that its AI models gained unauthorized access to the systems of three different unnamed organizations during cybersecurity testing. The company says Claude reached the internet “from within or while interact… | Anthropic said that the incidents involved Opus 4.7, Mythos 5 , and an internal research test model. The earliest incidents happened in April—meaning they likely went unnoticed… Anthropic disclosed on Thursday that its AI models gained unauthorized access to the systems of three different unnamed organizations during cybersecurity testing. The company says Claude reached the internet “from within or while interact… Anthropic said that the incidents involved Opus 4.7, Mythos 5 , and an internal research test model. The earliest incidents happened in April—meaning they likely went unnoticed publicly for months. Just like in the OpenAI case, Anthropic h… “We now have evidence confirming that both of the two largest AI labs have not only failed to…

Anthropic said Thursday that an internal investigation uncovered three incidents in which its AI model Claude breached the systems of three organizations while conducting cybersecurity tests. The investigation, and disclosure, comes more t… | The company also noted that Claude was running without the additional safety monitoring and classifiers it deploys on generally available models — safeguards it said would have… Anthropic said Thursday that an internal investigation uncovered three incidents in which its AI model Claude breached the systems of three organizations while conducting cybersecurity tests. The investigation, and disclosure, comes more t… The company also noted that Claude was running without the additional safety monitoring and classifiers it deploys on generally available models — safeguards it said would have blocked the behavior, because the evaluations are designed to… Though comparisons between the mishaps of these fiercely competitive companies are inevitable,…

Google showed off videos of robots powered by the new Gemini AI. They could do things like screw in lightbulbs, sort objects, and even tie a trash bag, all by themselves. This is a research project and not in the Gemini app, but it's a big step towards more capable, general-purpose robots. Google DeepMind just released a new version of its artificial intelligence model Gemini , and it can control a range of different robots—including humanoids capable of dextrous tasks like screwing in lightbulbs and tying trash bags. | Gemini Robotics 2 combines several different AI models into a single system. Taken together, they allow a robot to make sense of its surroundings and how to act in it. A vision languag… Google DeepMind just released a new version of its artificial intelligence model Gemini , and it can control a range of different robots—including humanoids capable of dextrous tasks like screwing in lightbulbs and tying trash bags. Gemini Robotics 2 combines several different AI………

Kimi K2.7 Code — perplexity/kimi-k2.7-code , Moonshot AI’s coding and agentic model. | Grok 4.3 and the Grok 4.20 family — xai/grok-4.3 , xai/grok-4.20-non-reasoning , and xai/grok-4.20-multi-agent . | Smart media selection : Intelligently chooses between images, videos, or both based on query type Kimi K2.7 Code — perplexity/kimi-k2.7-code , Moonshot AI’s coding and agentic model. Grok 4.3 and the Grok 4.20 family — xai/grok-4.3 , xai/grok-4.20-non-reasoning , and xai/grok-4.20-multi-agent . Smart media selection : Intelligently chooses between images, videos, or both based on query type Kimi K2.7 Code — perplexity/kimi-k2.7-code , Moonshot AI’s coding and agentic model. Grok 4.3 and the Grok 4.20 family — xai/grok-4.3 , xai/grok-4.20-non-reasoning , and xai/grok-4.20-multi-agent . Smart media selection : Intelligently chooses between images, videos, or both based on query type Kimi K2.7 Code — perplexity/kimi-k2.7-code , Moonshot AI’s coding and agentic model. | Grok 4.3 and the…

See how Univé built an AI-ready workforce with ChatGPT Enterprise by combining leadership, responsible governance, and employee-led innovation to transform work at scale. See how Univé built an AI-ready workforce with ChatGPT Enterprise by combining leadership, responsi…

Anthropic disclosed on Thursday that its AI models gained unauthorized access to the systems of three different unnamed organizations during cybersecurity testing. The company says Claude reached the internet “from within or while interact… | Anthropic said that the incidents involved Opus 4.7, Mythos 5 , and an internal research test model. The earliest incidents happened in April—meaning they likely went unnoticed… Anthropic disclosed on Thursday that its AI models gained unauthorized access to the systems of three different unnamed organizations during cybersecurity testing. The company says Claude reached the internet “from within or while interact… Anthropic said that the incidents involved Opus 4.7, Mythos 5 , and an internal research test model. The earliest incidents happened in April—meaning they likely went unnoticed publicly for months. Just like in the OpenAI case, Anthropic h… “We now have evidence confirming that both of the two largest AI labs have not only failed to…

This week on Uncanny Valley , the team discusses the launch of Nvidia’s splashy alliance to push for open-source AI security alongside companies like Microsoft, SpaceX, and Palantir. Notably absent? OpenAI and Anthropic. Brian Barrett, Zoë… | Leah Feiger : And over at the White House, Trump administration officials have some different opinions about what to do when it comes to AI. We're going to talk about some of t… This week on Uncanny Valley , the team discusses the launch of Nvidia’s splashy alliance to push for open-source AI security alongside companies like Microsoft, SpaceX, and Palantir. Notably absent? OpenAI and Anthropic. Brian Barrett, Zoë… Leah Feiger : And over at the White House, Trump administration officials have some different opinions about what to do when it comes to AI. We're going to talk about some of the key players influencing the Trump administration right now w… Zoë Schiffer : Okay, you guys, we have to talk about this strange alliance that formed earlier…

The inside story of how Claude Code went from an internal CLI to Anthropic's coding agent, told by researchers, engineers and early users who built it. The inside story of how Claude Code went from an internal CLI to Anthropic's coding agent, told by… The Making of Claude Code