Goodfire says its new ‘inside-out’ monitors catch rogue AI agents at a fraction of the cost

Key points

• Goodfire has launched monitors that track AI models internally, reducing costs by reusing existing computations. The system, available to Baseten customers, detects risks like hacking and misuse.
• Goodfire claims its approach is 94% effective at catching malicious activity, with minimal impact on response times.
• The startup aims to address vulnerabilities in open AI models, which lack the safeguards of closed systems.