Public entity risk pools · BY INVITATION
Where the pool stands.Before AI answers for its members.
AI already reads what each member publishes and answers on its behalf, accurately or not. Misinformation is the leading AI concern, and the cyber threat that travels with it has moved into the news. Both begin in the same place: the websites and documents a member publishes, including the many built in the rush online during COVID that remain live, looked after by small teams with little room to take on more.
The briefing is 20 minutes, private and by invitation, and brings 2 questions into the open for the pool: is the membership fit for AI visibility, and are directors and officers, at the member and at the pool, becoming increasingly, and unnecessarily, exposed as AI grows?
Recent weeks
Members are more vulnerable to cyber attack, and this week showed why.
On 21 Oct. 2026 governments in Finland, the Netherlands, Germany, Belgium and elsewhere took their own public websites offline. Two flaws had been found in PHP, the software behind close to 70% of the world’s websites, that let an attacker run code on a web server without a password. The US National Security Agency told organisations to patch at once or take their servers down; hospitals, schools and colleges were among those that went dark; AI found the weakness, and AI is what lets attackers turn a weakness like it into a working attack in hours rather than weeks.
That lands hardest on members. A charter school, a special district or a small municipality is more likely to run its websites on lower-end platforms built on exactly this software, with limited vendor support, looked after by 1 or 2 people alongside other duties, and to have kept the sites it set up quickly during COVID. Updates arrive late or not at all, and nobody is watching the older sites. Those are the conditions an automated attack looks for, and the reason an organisation once too small to be worth the effort is now a target.
For a pool the question is how many members are exposed at the same time, and whether anyone can see it. The week’s events put that question to risk leadership directly (CNBC, 22 Oct. 2026cnbc.comWe do not enable external links - please copy and paste, you are then certain as to the link being opened).
The leading concern
Misinformation leads the list, and it starts at home.
Across businesses surveyed by Gallagher in 2026, 57% name AI errors, misinformation and hallucinations as the leading AI threat, ahead of any other. AAAnow research finds that in 79% of misinformation or misrepresentation, the organisation’s own online position is the root cause. And 41% of the typical web estate is unknown to the organisation that owns it.
Gallagher, 2026. AI Adoption and Risk Benchmarking. AAAnow Research, Dec. 2023 to Jan. 2026; P&C (human misinformation) 2014 to 2023. P&C / Sitemorse risk profiling, 2017 to 2023, covering more than 119 million websites.
The concern is widely put down to the technology. The evidence places most of it with the information AI is given to read: outdated, inconsistent, incomplete or hard to interpret. That is where it can be measured, and where a member, with the pool behind it, can act.
At pool level
Beyond one member, one team: why the pool carries it.
The exposure begins inside a member and is carried by the pool. A wrong AI answer about a school district or a special district is read by parents, press and the public as the organisation’s own statement, and the claim, the incident response and the question of oversight arrive at the pool. The briefing examines the issue at pool level without assuming that all members face the same exposure, or that each exposure results in a claim.
The first step
One approach across the pool, and evidence the pool can show.
The service is deployed by the pool and runs the same way for each member. The 4 stages below set out what the pool gets from it: a consistent view of AI exposure across the membership, members able to start reducing their risk, and auditing in place that the pool can show to those providing its coverage.
Together · request with the form alongsiderequest with the form below
Stage 1. The briefing.
Twenty minutes for pool executive and risk leadership on the evidence and the method, then open discussion for as long as the questions run. Held privately, by invitation, on a call we confirm personally.
AAAnow
Stage 2. The audit across the pool.
Each member’s key websites and assets are read the same way from the outside, as AI reads them, and graded against the AI Readiness maturity scale, benchmarking the foundations first. The pool holds 1 consistent, dated picture of where its membership stands on AI exposure, member by member, rather than a set of unconnected impressions.
Members
Stage 3. Members start to mitigate.
Each member receives its own position from the same audit, with what is driving its exposure and where to start. A charter school and a large school district see different results, read the same way, so the pool sets its guidance and priorities on one basis and members begin reducing their risk.
The pool
Stage 4. Evidence for those providing the coverage.
Repeated, the audit becomes the pool’s ongoing view of its membership’s exposure, and a record that a consistent level of auditing is in place. The pool can show its carriers and reinsurers that AI exposure across the membership is measured, watched and being mitigated.
In the room
What is said in the room stays in the room.
The call spends 20 minutes on evidence and method, then opens to discussion. The material is prepared for the call and kept to limited distribution, so the conversation about the membership can be direct.
The Insurance Pools AI Readiness Benchmark shows how pools themselves read from the outside.
Commonly asked questions.
What is this briefing about?
The briefing addresses 2 connected but distinct exposures affecting public entity pools: rapidly escalating AI-enabled cyber exposure, and AI misinformation or misrepresentation.
Why is this relevant to a public entity risk pool?
Member exposures may create wider considerations involving shared risk control, governance, claims handling, incident response, and the support provided across the pool’s membership.
Why does this issue require attention now?
Artificial intelligence is increasing the speed, scale, accessibility, and repeatability of cyber threats and misinformation, making both exposures harder to understand and control.
What does Gallagher’s 57% finding demonstrate?
It shows that AI errors, misinformation and hallucinations are already the market’s leading perceived AI concern, rather than an isolated or theoretical technology issue. Gallagher, 2026. AI Adoption and Risk Benchmarking.
How can an organisation’s own information contribute to misinformation?
AI systems rely on published information, which can be misread or repeated when that information is outdated, conflicting, incomplete, unclear, or poorly governed.
Can AI misinformation genuinely be measured?
Yes. AI misinformation exposure can be assessed systematically and evidenced, rather than relying solely on general visibility measures, individual examples, or subjective opinion.
Can the exposure be controlled?
Significant causes can be identified and addressed through stronger information control, governance, measurement, and ongoing risk management, although not all elements of risk can be eliminated.
Is this the same as a general AI Readiness briefing?
No. This session is scoped specifically to the 2 exposures that matter most for public entity pools. For the fuller picture, outside-in discovery, governance and visibility, the main briefing covers the full discipline.
Has the pool or its membership already been assessed?
No. The briefing comes first. The audit across the pool is the second stage, and follows when the pool chooses to take it.
Why is AAAnow qualified to deliver this briefing?
AAAnow is part of the group behind Sitemorse®, Privacy and Cookies, ARWELL and AAAtraq®: over 25 years of unbiased outside-in digital assessment, including some of the earliest AI risk profiling for accessibility compliance. More on our experience is on the About page.
Sources
- Gallagher, 2026. AI Adoption and Risk Benchmarking. Businesses cite AI errors, misinformation and hallucinations as the leading AI threat: 57%. ajg.comGallagherWe do not enable external links - please copy and paste, you are then certain as to the link being opened
- AAAnow Research, Dec. 2023 to Jan. 2026; P&C (human misinformation) 2014 to 2023. In 79% of misinformation or misrepresentation, the organisation’s own online position is the root cause. aaanow.ai
- P&C / Sitemorse risk profiling, 2017 to 2023, covering more than 119 million websites. The share of the typical web estate unknown to the organisation: 41%. sitemorse.com
- CNBC, 22 Oct. 2026. Why governments pulled websites offline over the decades-old PHP programming language. The PHP2Shell flaws, the NSA warning, and websites taken offline in several countries on 21 Oct. 2026. cnbc.comCNBCWe do not enable external links - please copy and paste, you are then certain as to the link being opened
- Kaspersky, Oct. 2026. Critical PHP vulnerabilities put millions of websites worldwide at risk. PHP behind nearly 70% of websites; shutdowns affecting hospitals, educational institutions and government services; millions of exploit attempts recorded by Cloudflare. kaspersky.comKasperskyWe do not enable external links - please copy and paste, you are then certain as to the link being opened