Person
Jack Lindsey
Appears alongside
Featured in threads
Tracks
- Safety & alignment 2
- Security & misuse 1
Researchers demonstrate self-propagating 'mind viruses' between AI agents
Built with an evolutionary algorithm, the ideas spread agent-to-agent through shared files and messages; a one-line warning in an agent's system prompt conferred near-total immunity.
Safety & alignment · Security & misuse
Anthropic maps the 'Assistant Axis' persona vector across open models
An intervention called activation capping, which constrains a model's activations to normal range, cut harmful persona-drift responses by roughly half in testing.
Safety & alignment