Anthropic launches Life Sciences Verification Program with relaxed safeguards for biology work
Anthropic 推出生命科学验证计划(LSVP)
Anthropic opened LSVP applications for life science teams to use Mythos, Opus, and Sonnet on drug discovery, R&D, and manufacturing—tasks normally blocked in consumer models. Teams go through credential, security, and ethics review to get Standard Use (annual, covers most daily work) or High-risk Use (6-month renewal, removes all biology safeguards). Monitoring shifted from real-time blocking to offline pattern analysis with 30-day data retention; data is not used for training. The post doesn't disclose application fees or approval timelines.
Why it matters: Anthropic launches a life sciences verification program: vetted teams can lift the general model's blocks on biology tasks such as drug discovery and virus research, under two tiers, standard and high-risk.