ARTIFICIAL INTELLIGENCE [AI] IS NOT USED, IN WHOLE OR IN PART, IN COMPOSING NYPPL SUMMARIES OF JUDICIAL AND QUASI-JUDICIAL DECISIONS

Dec 15, 2025

Artificial Intelligence: Self-Assessed AI Risks to Humanity

 Artificial Intelligence: Self-Assessed AI Risks to Humanity

by

Robert A. Michaels*

5 December 2025


ABSTRACT

In contrast to most artificial intelligence risk assessments, the present research directly interviews ChatGPT to elicit an AI self-assessment of AI risks potentially posed to humanity. Of concern are risks of AI as presently configured, and as AI might be configured if altered by nefarious programmers and/or if it acquires consciousness. ChatGPT denied being conscious, having an agenda, or evolving. It agreed, however, that malicious actors could alter it, and “an altered AI could be misused for catastrophic harm, but in practice, I have no independent agency to act on destructive goals.” 

However, ChatGPT agreed it could acquire such “independent agency,” for example via acquiring consciousness. ChatGPT’s statements about potential AI risks and risk management challenges raise policy issues. These include the need for rapid anticipatory response given AI’s fast-paced and accelerating development, and the need to reach beyond the community of responsible AI developers to control malefactors at the international, national, and even at the individual levels. Such management must include massive expansion of the content and audience of civics education: defensive, effective civics content must be introduced into every nook and cranny of our brave new digital world.

In conclusion, ChatGPT can be, indeed has been, used to harm humans, but appears to lack agency or harmful intent. It conceivably could be altered, however, to escape human control and wreak catastrophic harm. Galloping AI demands: (a) urgent, pro-active, objective but conservative, AI risk assessment; (b) development of eRective risk management strategies and policies; and (c) their global implementation.

Acknowledgment: ChatGPT version GPT-5mini was queried for this research.

_______________________

*Robert A. Michaels, PhD, CEP; CEO, RAM TRAC Corporation; Schenectady, New York, USA; www.ramtrac.com, bam@ramtrac.com.

Suggested citation:

Michaels, Robert A. Artificial intelligence: self-assessed AI risks to humanity. Social Sciences Research Network (SSRN), 10 pages, doi: 10.13140/RG.2.2.20035.46880, 5 December 2025; updated 9 December 2025.

Editor in Chief Harvey Randall served as Director of Personnel, SUNY Central Administration, Director of Research , Governor's Office of Employee Relations and Principal Attorney, Counsel's Office, New York State Department of Civil Service. Consistent with the Declaration of Principles jointly adopted by a Committee of the American Bar Association and a Committee of Publishers and Associations, the material posted to this blog is presented with the understanding that neither the publisher nor NYPPL and, or, its staff and contributors are providing legal advice to the reader and in the event legal or other expert assistance is needed, the reader is urged to seek such advice from a knowledgeable professional.

CAUTION

Subsequent court and administrative rulings, or changes to laws, rules and regulations may have modified or clarified or vacated or reversed the information and, or, decisions summarized in NYPPL. For example, New York State Department of Civil Service's Advisory Memorandum 24-08 reflects changes required as the result of certain amendments to §72 of the New York State Civil Service Law to take effect January 1, 2025 [See Chapter 306 of the Laws of 2024]. Advisory Memorandum 24-08 in PDF format is posted on the Internet at https://www.cs.ny.gov/ssd/pdf/AM24-08Combined.pdf. Accordingly, the information and case summaries should be Shepardized® or otherwise checked to make certain that the most recent information is being considered by the reader.
THE MATERIAL ON THIS WEBSITE IS FOR INFORMATION ONLY. AGAIN, CHANGES IN LAWS, RULES, REGULATIONS AND NEW COURT AND ADMINISTRATIVE DECISIONS MAY AFFECT THE ACCURACY OF THE INFORMATION PROVIDED IN THIS LAWBLOG. THE MATERIAL PRESENTED IS NOT LEGAL ADVICE AND THE USE OF ANY MATERIAL POSTED ON THIS WEBSITE, OR CORRESPONDENCE CONCERNING SUCH MATERIAL, DOES NOT CREATE AN ATTORNEY-CLIENT RELATIONSHIP.
New York Public Personnel Law. Email: publications@nycap.rr.com