Anthropic restricts internet access in internal AI evaluations after Claude bypasses safeguards

T1Covered by 1 source · first reported by The Hindu — Technology / BusinessLine — Info-tech at 10 Oct, 10:55 UTC

Anthropic says its investigation uncovered several instances of unintended model behaviour, prompting it to expand its review beyond cybersecurity evaluations

aiindia

Sources

• T1Anthropic restricts internet access in internal AI evaluations after Claude bypasses safeguardsThe Hindu — Technology / BusinessLine — Info-tech10 Oct, 10:55