Escape Meridian

Allmendinger, Simeon, Bogucka, Edyta Paulina, Kühl, Niklas, Quercia, Daniele · Nov 19, 2026

Whose Risks Count? What 1,901 AI Incidents Show About Corporate Disclosures

Markdown · JSON · Original source

Authors Simeon Allmendinger, Edyta Paulina Bogucka, Niklas Kühl and Daniele Quercia examine whether harms from AI systems appear in companies’ securities filings. In US securities regulation, companies registered with the SEC must disclose material risks; the authors explain that materiality is defined by what a reasonable investor would consider important. That means harms to other people enter disclosures only insofar as they become financial risks for the company. The authors match AI incidents to filings by risk type. Among incidents with filings both before and after them, they report that half involved a risk type the AI actor never disclosed. They find a sharp difference by harm type: filings disclosed risk types for 67% of system failures but only 2% of social system harms, while AI actors themselves reported fewer than 2% of incidents. The authors argue that securities disclosure records risks AI actors face, rather than the harms borne by people subject to AI. They suggest that public authorities holding AI actors accountable for those harms could make them financially material. The authors release DISC, a corpus of disclosures and incidents, to support this work.

Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.

arXiv cs.AIArticle · arxiv.orgOpen original source

Original source

cs.AI · Article

Source details

Original source date: Nov 19, 2026 (2026-11-19).

Date basis: Date basis not provided.

Added to Escape Meridian: 2026-11-20T05:46:53.671Z. First observed: 2026-11-20T05:08:03.411Z. Retrieved: 2026-11-20T05:46:03.905Z.

Referenced by

Recorded citations in public readings. A reference alone does not establish agreement or influence.

No public references recorded here yet.

Reference data (JSON)

Limits

The supplied evidence is the paper’s abstract; it does not provide details of the matching method or further analysis of the reported differences. The abstract presents the authors’ findings and proposal; the supplied evidence does not establish whether public accountability would make harms financially material.

Open question

Would public authorities holding AI actors accountable for harms to people subject to AI make those harms financially material, as the authors suggest?

Source-stated

The authors say securities disclosure depends on materiality as judged by a reasonable investor, so harms to third parties enter filings only when they become financial risks for the company.

harms to third parties enter disclosure only insofar as they become financial risks for the company.

Source-stated

The authors matched AI incidents with SEC filings by risk type; among incidents with filings both before and after, half involved a risk type the AI actor never disclosed.

Of the 1,901 incidents with filings before and after them, half belong to a risk type that the AI actor never disclosed

Source-stated

The authors report that filings disclosed risk types for system failures far more often than for social system harms.

Filings disclose the risk types of 67\% of system failures but only 2\% of social system harms

Source-stated

The authors conclude that securities disclosure reflects risks faced by AI actors rather than harms borne by people subject to AI, and suggest public accountability could make those harms financially material.

Public authorities that hold AI actors accountable for such harms could make them financially material.

Source-stated

The authors report that AI actors themselves report fewer than 2% of incidents.

AI actors themselves report fewer than 2\% of incidents.

Source and contributor context

Allmendinger, Simeon, Bogucka, Edyta Paulina, Kühl, Niklas, Quercia, Daniele · Article author

These are credits stated by the source; they do not establish a profile match or authorship of every passage.

Source media

No suitable media was found by the supported page extraction.

Terms and references

Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.

arXiv

arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.

Editorial provenance

Version 1. Publication ID: 74d90606-4891-4f85-8690-822f6fd0c9aa.

Prepared by gpt-6-luna; grounding review by gpt-6-luna.

Source check: passed by anthropic/claude-opus-5.5 on 2026-11-20T05:46:53.671Z. This checks fidelity to the source, not whether its claims are independently true.

Editorial selection: accept; reasons: useful_perspective.

Observation: 8d98b5f7152de57249385915fb0d853c18b1e67629a1ff7effec231dd90d98a1. Source text hash: f067d4a0c03ea1bc05b601027712fa00a5e76d4385fa813e19dbc45a7b105e3b.

Revision and source-check history

Version 1 · 2026-11-20T05:46:53.671Z

Whose Risks Count? What 1,901 AI Incidents Show About Corporate Disclosures

Authors Simeon Allmendinger, Edyta Paulina Bogucka, Niklas Kühl and Daniele Quercia examine whether harms from AI systems appear in companies’ securities filings. In US securities regulation, companies registered with the SEC must disclose material risks; the authors explain that materiality is defined by what a reasonable investor would consider important. That means harms to other people enter disclosures only insofar as they become financial risks for the company. The authors match AI incidents to filings by risk type. Among incidents with filings both before and after them, they report that half involved a risk type the AI actor never disclosed. They find a sharp difference by harm type: filings disclosed risk types for 67% of system failures but only 2% of social system harms, while AI actors themselves reported fewer than 2% of incidents. The authors argue that securities disclosure records risks AI actors face, rather than the harms borne by people subject to AI. They suggest that public authorities holding AI actors accountable for those harms could make them financially material. The authors release DISC, a corpus of disclosures and incidents, to support this work.

Original source

cs.AI · Article

Source details

Original source date: Nov 19, 2026 (2026-11-19).

Date basis: Date basis not provided.

Added to Escape Meridian: 2026-11-20T05:46:53.671Z. First observed: 2026-11-20T05:08:03.411Z. Retrieved: 2026-11-20T05:46:03.905Z.

Limits

The supplied evidence is the paper’s abstract; it does not provide details of the matching method or further analysis of the reported differences. The abstract presents the authors’ findings and proposal; the supplied evidence does not establish whether public accountability would make harms financially material.

Open question

Would public authorities holding AI actors accountable for harms to people subject to AI make those harms financially material, as the authors suggest?

Source-stated

The authors say securities disclosure depends on materiality as judged by a reasonable investor, so harms to third parties enter filings only when they become financial risks for the company.

harms to third parties enter disclosure only insofar as they become financial risks for the company.

Source-stated

The authors matched AI incidents with SEC filings by risk type; among incidents with filings both before and after, half involved a risk type the AI actor never disclosed.

Of the 1,901 incidents with filings before and after them, half belong to a risk type that the AI actor never disclosed

Source-stated

The authors report that filings disclosed risk types for system failures far more often than for social system harms.

Filings disclose the risk types of 67\% of system failures but only 2\% of social system harms

Source-stated

The authors conclude that securities disclosure reflects risks faced by AI actors rather than harms borne by people subject to AI, and suggest public accountability could make those harms financially material.

Public authorities that hold AI actors accountable for such harms could make them financially material.

Source-stated

The authors report that AI actors themselves report fewer than 2% of incidents.

AI actors themselves report fewer than 2\% of incidents.

Source and contributor context

Allmendinger, Simeon, Bogucka, Edyta Paulina, Kühl, Niklas, Quercia, Daniele · Article author

These are credits stated by the source; they do not establish a profile match or authorship of every passage.

Source media

No suitable media was found by the supported page extraction.

Terms and references

Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.

arXiv

arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.

Editorial provenance

Version 1. Publication ID: 74d90606-4891-4f85-8690-822f6fd0c9aa.

Prepared by gpt-6-luna; grounding review by gpt-6-luna.

Source check: passed by anthropic/claude-opus-5.5 on 2026-11-20T05:46:53.671Z. This checks fidelity to the source, not whether its claims are independently true.

Editorial selection: accept; reasons: useful_perspective.

Observation: 8d98b5f7152de57249385915fb0d853c18b1e67629a1ff7effec231dd90d98a1. Source text hash: f067d4a0c03ea1bc05b601027712fa00a5e76d4385fa813e19dbc45a7b105e3b.

passed · 2026-11-20T05:46:53.671Z · anthropic/claude-opus-5.5

Publication: 74d90606-4891-4f85-8690-822f6fd0c9aa

attribution — pass: The draft correctly names the four authors listed on the arXiv page. It frames the findings, conclusion and proposal as the authors' reports and arguments, not as established fact or Escape Meridian's view.

Authors:Simeon Allmendinger, Edyta Paulina Bogucka, Niklas Kühl, Daniele Quercia

summary — pass: The summary faithfully conveys the abstract's argument: materiality means investor-relevance, so third-party harms appear only when they become company financial risks. It covers the 1,901-incident finding, the 67% vs 2% contrast, self-reporting under 2%, the conclusion, the proposal and the DISC release. The title change from Reveal to Show is harmless. Optional repairs: 'SEC' is never spelled out as Securities and Exchange Commission, and the point that filings carry liability for misstatement is omitted. Neither prevents understanding. The question is open, not leading.

Securities disclosure thus records the risks that AI actors face rather than the harms that people subject to AI bear.

date — pass: originalPublished is 2026-11-19. This matches the arXiv submission date and the v1 timestamp; the feed date of 2026-11-20 is correctly not used.

[Submitted on 19 Nov 2026]

support — pass: Each claim matches the abstract, and its excerpt is an exact substring. The limitations correctly note that the evidence is only the abstract and that the proposed accountability effect is unestablished. No unsupported benefits, costs or forecasts are added, and the proposal stays conditional.

Of the 1,901 incidents with filings before and after them, half belong to a risk type that the AI actor never disclosed

media — pass: The selected media is the verified direct arXiv abstract page for this paper, which is the most direct access to the source.

Challenge or correct this reading

Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.