6 canonical risk pages
Legal
Regulatory, compliance, rights, and liability exposure.
Automated Libel
Generation of false and defamatory information about real individuals via model hallucinations, with potential for serious reputational and legal damage.
Consent Violation
Use of personal, intimate, or sensitive data for model training without explicit informed consent from the affected individuals.
Copyright Violation
Unauthorized use of copyright-protected works in training datasets without rightsholder consent, generating legal controversies over intellectual property.
Identity Theft
Unauthorized commercial appropriation and use of individuals' image, voice, or personality via AI synthesis without compensation or consent.
Legal Liability
Absence of clear legal frameworks for attribution of civil and criminal liability when autonomous AI systems cause damages or errors with material consequences.
Illegal Scraping
Massive extraction of data from websites for model training ignoring robots.txt, terms of service, and data property rights.