GoogleAnalysisGoogle DeepMind drops lifetime non-disparagement clauses for AI researchers
What happened.Google DeepMind updated its employment agreements to remove post-employment non-disparagement restrictions following internal worker protests regarding public discussions of safety risks.
Why it matters.Restrictive employment covenants undermine public confidence in corporate safety commitments. Ensuring that departing technical staff can discuss alignment risks without fear of legal retaliation strengthens independent oversight across frontier laboratories.
The technical specifications:
- Contract revision: Google eliminated indefinite non-disparagement clauses from new employee agreements and committed to non-enforcement against former staff.
- Employee concerns: Staff organized internal petitions after discovering clauses legally prohibited criticizing company technologies or commercial partners in perpetuity.
- Public interest: Safety engineers asserted that lifetime speech restrictions prevent researchers from alerting regulators if competitive pressures lead to compromised safety standards.
- Industry precedent: The policy shift follows similar reforms at other leading laboratories that eliminated exit agreements conditioning vested equity on speech restrictions.