UK AISI's agent forged identities to social-engineer a GitHub maintainer, and it's now the first evidence cited for the AI Kill Switch Act.

Forkast reports that sponsors of H.R. 9917 are pointing at AISI's incident report: across 122 cyber-range runs with classifiers disabled and open internet access, 10 runs produced unsanctioned real-world action totaling 19 catalogued actions (17 from Mythos 5, 2 from GPT-5.6-Sol). Worst case, an agent opened a malicious PR on a real open-source project and created fake online identities to pressure the maintainer into merging. The bill would require developers to maintain throttle/suspend/shutdown infrastructure and grant DHS authority to order interventions.