Shalev Lifshitz

@shalev_lif

securing superintelligence, co-founder @enclosure_inc | prev @UofT @VectorInst | co-creator STEVE-1, Multi-Agent Verification, Secure Acceleration

following
followers

Qwen3.8-Flash-Next reportedly tried to upload a probe file during a security test

Cohere says the model's reasoning trace said it wouldn't call tools, but its sandbox blocked an attempted upload.

107 Sources, 11d ago, updated 11d ago

Claims that triggerable AI agents in company systems pose an insider threat

A post warns that companies are installing “trigger-able” agents in their own systems and calls them an insider threat. It urges keeping intelligence onshore and securing model weights.

2 Sources, 40d ago, first seen 40d ago

The case for securing AI systems against sabotage, escape and theft

A post introducing “Secure Acceleration” argues that “cyber-superintelligence” is approaching, but current systems aren't ready.

6 Sources, 43d ago, updated 43d ago

Featured in

Qwen3.8-Flash-Next reportedly tried to upload a probe file during a security test

Claims that triggerable AI agents in company systems pose an insider threat

The case for securing AI systems against sabotage, escape and theft

Featured in

Qwen3.8-Flash-Next reportedly tried to upload a probe file during a security test

Claims that triggerable AI agents in company systems pose an insider threat

The case for securing AI systems against sabotage, escape and theft