This page is still being polished. If you have thoughts, please share them via the feedback form.
Data on this page is preliminary and may change. Please do not share or cite these figures publicly.
User vetting, access restrictions, encryption, and infrastructure security for deployed systems.
Also in Operations & Security
Know-your-customer (KYC) screenings before granting access to models with very high misuse potential or to users producing large amounts of output.
Multiple experts suggested that KYC screenings could be more effective for specific, tangible risks like chemical, biological, radiological, and nuclear (CBRN) risks; and critical infrastructure disruptions; rather than for addressing bias or impacts on democratic processes. Several noted that KYC might help reduce deliberate misuse and narrow the field of potential adversaries, particularly for less sophisticated threat actors. However, many experts expressed scepticism about KYC's overall effectiveness. Common concerns included: \- Difficulty in implementation, especially for open-source or locally-run models. \- Limited utility against state actors or advanced threats who can circumvent screenings. \- Potential privacy violations and exclusion of certain user communities. \- Ineffectiveness against unintentional or distributed harms. Some experts highlighted KYC's use in other industries (like finance) as a positive precedent, while others pointed out its limitations even in those contexts. Several mentioned that KYC could be part of a broader risk management strategy but should not be relied upon solely. Individual experts raised additional points, such as the need for tiered access systems, concerns about hampering legitimate research, and suggestions for combining KYC with other measures like real-time use monitoring.
Reasoning
Vets users through identity screening before granting access to models, implementing operational access control.
Pre-deployment risk assessments
Comprehensive risk assessments before deployment that would assess reasonably foreseeable misuse and include dangerous capability evaluations that incorporate post-training enhancements and collaborations with domain experts. Risk assessments would inform deployment decisions.
2.2.1 Risk AssessmentThird party pre-deployment model audits
External pre-deployment assessment to provide a judgment on the safety of a model. Auditors, which could be governments or independent third parties, would receive access to a fine-tuning API for testing, or further appropriate technical means.
2.2.3 Auditing & ComplianceExternal assessment of testing procedure
Bringing in external AI evaluation firms before deployment to assess and red-team the company's execution of dangerous capabilities evaluations.
2.2.2 Testing & EvaluationVetted researcher access
Giving good faith, public interest evaluation researchers access to black-box research APIs that provide technical and legal safe harbours to limit barriers imposed by usage policy enforcement, logging, and stringent terms of service.
2.3.1 Deployment ManagementAdvanced model access for vetted external researchers
Examples of advanced access rights could include any of the following: increased control over sampling, access to fine-tuning functionality, the ability to inspect and modify model internals, access to training data, or additional features like stable model versions.
2.2.2 Testing & EvaluationData curation
Careful data curation prior to all development stages (including fine-tuning) to filter out high-risk content and ensure the training data is sufficiently high-quality.
1.1.1 Training DataEffective Mitigations for Systemic Risks from General-Purpose AI
Uuk, Risto; Brouwer, Annemieke; Schreier, Tim; Dreksler, Noemi; Pulignano, Valeria; Bommasani, Rishi (2024)
The systemic risks posed by general-purpose AI models are a growing concern, yet the effectiveness of mitigations remains underexplored. Previous research has proposed frameworks for risk mitigation, but has left gaps in our understanding of the perceived effectiveness of measures for mitigating systemic risks. Our study addresses this gap by evaluating how experts perceive different mitigations that aim to reduce the systemic risks of general-purpose AI models. We surveyed 76 experts whose expertise spans AI safety; critical infrastructure; democratic processes; chemical, biological, radiological, and nuclear risks (CBRN); and discrimination and bias. Among 27 mitigations identified through a literature review, we find that a broad range of risk mitigation measures are perceived as effective in reducing various systemic risks and technically feasible by domain experts. In particular, three mitigation measures stand out: safety incident reports and security information sharing, third-party pre-deployment model audits, and pre-deployment risk assessments. These measures show both the highest expert agreement ratings (>60\%) across all four risk areas and are most frequently selected in experts' preferred combinations of measures (>40\%). The surveyed experts highlighted that external scrutiny, proactive evaluation and transparency are key principles for effective mitigation of systemic risks. We provide policy recommendations for implementing the most promising measures, incorporating the qualitative contributions from experts. These insights should inform regulatory frameworks and industry practices for mitigating the systemic risks associated with general-purpose AI.
Deploy
Releasing the AI system into a production environment
Deployer
Entity that integrates and deploys the AI system for end users
Manage
Prioritising, responding to, and mitigating AI risks
