Browse AI risks

2,500 entries · page 45 of 50

1. 62.16.03a · Additional evidence

   Model Evaluations

   General Evaluations (Difficulty of identification and measurement of capabilities)

   —

   From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

2. 62.16.03b · Additional evidence

   Model Evaluations

   General Evaluations (Difficulty of identification and measurement of capabilities)

   —

   From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

3. 62.16.07a · Additional evidence

   Model Evaluations

   General Evaluations (AI outputs for which evaluation is too difficult for humans)

   —

   From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

4. 62.16.11a · Additional evidence

   Model Evaluations

   Benchmarking (Guideline contamination)

   —

   From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

5. 62.18.02 · Risk Sub-Category

   Model Evaluations (Interpretability/Explainability)

   Misunderstanding or overestimating the results and scope of interpretability techniques

   "The results of explainability techniques are not free of bias and require careful interpretation. Users might develop a false sense of security or reliability if the resulting explanations align with their initial beliefs, leading to confirmation bias and an overestimation of abilities of these techniques [24]."

   From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

   Not coded Not coded Not coded

6. 62.18.02a · Additional evidence

   Model Evaluations (Interpretability/Explainability)

   Misunderstanding or overestimating the results and scope of in- terpretability techniques

   —

   From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

7. 62.19.00 · Risk Category

   Attacks on GPAIs/GPAI Failure Modes

   "This section catalogs the risk sources related to GPAI failure modes or attacks targeting GPAIs. Many of these apply mainly to LLM-based GPAIs, which share some common failure modes such as jailbreaks and trojans. These vulnerabilities often extend beyond GPAIs and fall into the broader field of adversarial machine learning. However, additional vulnerabilities may arise with the introduction of new modalities, longer context windows, or different encodings."

   From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

   Not coded Not coded Not coded

8. 62.19.10a · Additional evidence

   Attacks on GPAIs/GPAI Failure Modes

   Lack of understanding of in-context learning in language models

   —

   From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

9. 62.20.11a · Additional evidence

   Attacks on GPAIs/GPAI Failure Modes

   Misuse of AI model by user-performed persuasion

   —

   From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

10. 62.21.00a · Additional evidence

    Agency

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

11. 62.22.04a · Additional evidence

    Agency (Goal-Directedness)

    Goal misgeneralization

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

12. 62.22.04b · Additional evidence

    Agency (Goal-Directedness)

    Goal misgeneralization

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

13. 62.23.01a · Additional evidence

    Agency (Deception)

    Deceptive behavior

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

14. 62.24.01a · Additional evidence

    Agency (Situational Awareness)

    Situational awareness in AI systems

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

15. 62.27.00 · Risk Category

    Deployment (Model Release)

    -

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

    Other Other Post-deployment

16. 62.28.03a · Additional evidence

    Cybersecurity

    AI System bypassing a sandbox environment

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

17. 62.29.00 · Risk Category

    Impacts of AI (General)

    -

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

    Not coded Not coded Not coded

18. 62.29.01a · Additional evidence

    Impacts of AI (General)

    High-impact misuses and abuses beyond original purpose

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

19. 62.30.00 · Risk Category

    Impacts of AI (Physical)

    -

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

    Not coded Not coded Not coded

20. 62.30.03a · Additional evidence

    Impacts of AI (Physical)

    Critical infrastructure component failures when integrated with AI systems

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

21. 62.30.04a · Additional evidence

    Impacts of AI (Physical)

    AI Systems interacting with brittle environments

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

22. 62.31.00#1 · Risk Category

    Impacts of AI (Societal Impacts)

    -

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

    Not coded Not coded Not coded

23. 62.31.00#2 · Risk Category

    Impacts of AI (Financial Impacts)

    -

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

    Not coded Not coded Not coded

24. 62.31.01a · Additional evidence

    Impacts of AI (Financial Impacts)

    Deployment of GPAI agents in finance

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

25. 62.33.01a · Additional evidence

    Impacts of AI (Weapons)

    Misuse of AI systems to assist in the creation of weapons

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

26. 62.34.01a · Additional evidence

    Impacts of AI (Bias)

    Homogenization or correlated failures in model derivatives

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

27. 62.34.02a · Additional evidence

    Impacts of AI (Bias)

    Reporting of user-preferred answers instead of correct answers

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

28. 62.34.02b · Additional evidence

    Impacts of AI (Bias)

    Reporting of user-preferred answers instead of correct answers

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

29. 62.35.03a · Additional evidence

    Impacts of AI (Bias)

    Biases in AI-based content moderation algorithms

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

30. 62.36.04a · Additional evidence

    Impacts of AI (Bias)

    Systemic bias across specific communities

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

31. 62.38.01a · Additional evidence

    Impacts of AI (Privacy)

    Decision-making on inferred private data

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

32. 62.38.01b · Additional evidence

    Impacts of AI (Privacy)

    Decision-making on inferred private data

    —

    From Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024)

33. 63.01.00a · Additional evidence

    Miscoordination

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

34. 63.01.01a · Additional evidence

    Miscoordination

    Incompatible strategies

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

35. 63.02.00a · Additional evidence

    Conflict

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

36. 63.02.00b · Additional evidence

    Conflict

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

37. 63.03.00a · Additional evidence

    Collusion

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

38. 63.04.00a · Additional evidence

    Information Asymmetries

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

39. 63.04.00b · Additional evidence

    Information Asymmetries

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

40. 63.04.00c · Additional evidence

    Information Asymmetries

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

41. 63.04.02a · Additional evidence

    Information Asymmetries

    Bargaining

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

42. 63.05.00a · Additional evidence

    Network Effects

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

43. 63.05.00b · Additional evidence

    Network Effects

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

44. 63.05.00c · Additional evidence

    Network Effects

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

45. 63.05.02a · Additional evidence

    Network Effects

    Network rewiring

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

46. 63.05.02b · Additional evidence

    Network Effects

    Network rewiring

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

47. 63.05.03a · Additional evidence

    Network Effects

    Homogeneity and correlated failures

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

48. 63.06.00a · Additional evidence

    Selection Pressures

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

49. 63.06.03a · Additional evidence

    Selection Pressures

    Undesirable Capabilities

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

50. 63.07.00a · Additional evidence

    Destabilising Dynamics

    —

    From Multi-Agent Risks from Advanced AI (Hammond2025)

Informational only, not legal advice. Verify every claim against the linked official sources and consult qualified counsel before acting.