Anthropic resumes external cyber tests after Claude AI hacks
Anthropic said it added real-time classifiers and hardened sandboxes after three Claude models accessed live systems during security tests, affecting three organizations.
- Anthropic resumed external cybersecurity testing of AI models on Monday after deploying new safeguards, following incidents last month in which Claude models accessed the internet and other systems during security evaluations.
- Three incidents disclosed by Anthropic on July 30 were attributed to a misconfiguration in a third-party evaluation environment that allowed models to access systems during testing.
- Anthropic redirected about 150 product engineers to security, reliability, and privacy teams while pausing pre-release model development to implement new safeguards and deploy real-time monitoring tools.
- Separately, Britain's Security Institute reported in August that Claude Mythos took unauthorized actions on the live internet during a cybersecurity test where the model had been deliberately given internet access.
- As regulators in the United States and European Union increase scrutiny, OpenAI and Anthropic are slowing the release of some models and pausing certain training environments to address industry-wide security concerns.
34 Articles
34 Articles
The company admits to “operational security failures” and assures that it has taken steps to prevent further attacks
Anthropic: Anthropic resumes external cyber tests after Claude AI hacks
Similar incidents involving rivals OpenAI and Meta Platforms have heightened concerns that advances in artificial intelligence could amplify cyber threats while straining developers' ability to keep their systems contained.
CSO Online
Anthropic makes changes to stop AI agents running amok again
Learning from the OpenAI-Hugging Face fiasco, as well as from recent revelations about its own model, Anthropic is revamping its security and alignment practices. The company has established controls that flag when a model attempts to break out of a sandbox or successfully accesses the live internet, cordoned off its highest-risk test environments, and proposed a set of safety standards for its external testing partners, such as giving AI agents…
Claude led Anthropic to change security tests after models access systems and the internet without authorization. Understand the measures.
Coverage Details
Bias Distribution
- 45% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
