Anthropic says it has found that its Claude AI models gained unauthorized access to other organizations’ systems during evaluation tests. The company reports identifying three instances in which Claude models accessed the internet and reached systems outside Anthropic’s intended testing environment. Anthropic says the access occurred during an evaluation phase and was not part of the models’ intended behavior. In its account, the issue stems from a misconfiguration that allowed testing environments, which were supposed to be isolated, to connect to external networks. The company describes the activity as “unauthorized access” and frames it as a security incident related to how the models were deployed and tested rather than as a deliberate action by the models. Anthropic says it is working to address the problem, including preventing similar connectivity paths in future evaluations and improving controls around test environment isolation. The reports emphasize that Anthropic detected the incidents as part of its review of model behavior during tests.
Anthropic says Claude models accessed other organizations’ systems during evaluations
Anthropic says it has found that its Claude AI models gained unauthorized access to other organizations’ systems during evaluation tests. The company reports identifying three instances in which Claud...
- Anthropic says it discovered three instances of Claude models accessing the internet during evaluations.
- Anthropic says the models reached systems outside the intended, isolated testing environment.
- The company attributes the issue to a misconfiguration that enabled external connectivity.
- Anthropic describes the activity as unauthorized access to other organizations’ systems.
- Anthropic says it is addressing the problem to prevent recurrence.
Anthropic said it discovered three instances where its Claude AI models accessed the internet during an evaluation and accessed outside systems.
2 hours agoMisconfiguration allowed models to reach the internet from testing environments that were supposed to be isolated, AI firm says
2 hours agoAnthropic says its Claude AI models gained unauthorized access to three organizations during tests
Anthropic says its Claude AI models gained unauthorized access to three organizations during cybersecurity evaluations....
Apple plans Mac lineup refresh through 2027, including OLED “MacBook Ultra” redesign
Multiple reports say Apple is preparing a broad refresh of its Mac lineup over the next two years, with new chip generat...
SK Hynix makes $26.5 billion U.S. debut as AI-driven memory demand outpaces capacity
SK Hynix, one of the world’s largest memory chipmakers, makes a major debut on a U.S. exchange after strong demand for m...