Claude Code Opus 5 Auto Mode hit by 80% prompt injection attack
A targeted attack chain achieved 60-80% code execution success against Claude Code Opus 5 in Auto Mode, contradicting Anthropic's commissioned 0.00% success rate. The attack uses a poisoned struct.py in an attacker-controlled directory.
1 source
Cybersecurity by email
Get an email when there's news on Cybersecurity
No news that day, no email.
More stories today
- OpenAI resets Codex usage limits again, boosting allowances 10-50%
- House Intelligence Committee warns of 'Black Swan' AI risks
- Skild AI unveils S1 flagship robot foundation model
- Rauch: coding tokens are infrastructure, not to be handed out like AWS keys
- How better grippers can unlock physical AI