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.
Red teaming, capability evaluations, adversarial testing, and performance verification.
Also in Risk & Assurance
Clear and user-friendly bug bounty programs that acknowledge and reward individuals for reporting model vulnerabilities and dangerous capabilities.
Multiple experts noted that bug bounty programs can be effective and have proven useful in other areas like cybersecurity. They are seen as a cost-effective way to leverage crowdsourced knowledge and promote accountability. Several experts mentioned that these programs could be helpful across various risk areas, particularly for identifying biases, discrimination, and vulnerabilities that could negatively affect democratic processes. However, there were also several concerns raised by multiple experts: \- The effectiveness may be limited for certain types of risks, especially those related to critical infrastructure or chemical, biological, radiological, and nuclear (CBRN) risks. \- There are potential information hazards, as the process might inadvertently teach people how to circumvent AI safeguards. \- The programs might not be as effective for subjective or value-laden issues. \- They occur post-deployment, which may be too late for some critical risks. Some individual experts raised specific points: \- One suggested that bug bounties could create a false sense of security if not properly implemented. \- Another mentioned that in certain contexts (like Brazil), such programs could be vulnerable to scams. \- One expert pointed out that these programs should not substitute robust, mandated risk assessments and expert auditing.
Reasoning
Crowdsourced vulnerability and capability discovery mechanism leverages external testing to identify model risks post-deployment.
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.
Operate and Monitor
Running, maintaining, and monitoring the AI system post-deployment
Developer
Entity that creates, trains, or modifies the AI system
Measure
Quantifying, testing, and monitoring identified AI risks
