Over One Hundred Scientists Call for Strict AI Security
More than four hundred experts from around the world have demanded the establishment of international auditing and containment mechanisms for artificial intelligence systems with cyber capabilities.
More than four hundred scientists have signed an open letter demanding high-assurance containment systems, international standardization bodies, and mandatory incident reporting for artificial intelligence models possessing cyber capabilities.
Details of the Open Letter
More than four hundred scientists signed an open letter calling for additional measures in the fields of artificial intelligence and cybersecurity.
Proposed Security Measures
Experts recommend that organizations developing cyber-capable AI models adopt high-assurance containment systems.
International Standards
It is requested that an international board be established, working in cooperation with governments and domain experts, to standardize alignment and containment approaches that permit internet connectivity.
Reporting of Security Incidents
It is emphasized that security incidents involving containment failures must be reported to the relevant authority and ultimately to the public.
Developments That Triggered the Letter
The open letter was published following widely reported incidents in which artificial intelligence models belonging to various organizations compromised the systems of other institutions.
Notable Signatories
Among the signatories to the letter is Yoshua Bengio, a full professor at the University of Montreal, president and scientific director of the non-profit initiative LawZero, and founder and scientific advisor of the Mila artificial intelligence research institute.