# goodfault

> goodfault is the neutral carrier for ai risk: insurance for ai agents, humanoid robots, drones, autonomous mobile robots, physical ai products, av software layers, and the startups that ship them. named perils written affirmatively, audit-gated underwriting, coverage quoted in minutes. tagline: every fault ends well. contact: hello@goodfault.com

goodfault writes the ai perils that standard policies are starting to exclude (the iso cg 40 47 generative-ai exclusions, effective january 1, 2026, attaching to renewals carrier by carrier). if a peril is named on our paper, it is covered.

## what we insure

- division a (bits): software agents. customer support, commerce, coding, operations, screening, and workflow agents.
- division b (atoms): physical ai. humanoid and general-purpose robots, autonomous mobile robots and industrial fleets, drones and uas, physical ai products, av and robotaxi programs.

## the named perils (division a, software agents)

- a1 unauthorized transactions: refunds, discounts, purchases, trades the agent was never authorized to make
- a2 hallucinated commitments: the agent invents a policy, price, or promise, and a court makes you honor it
- a3 data leakage by agent: pii exfiltration, cross-tenant exposure, ip disclosure caused by the agent's own actions
- a4 destructive actions: deleted records, wiped configs, broken pipelines, damaged production systems
- a5 injection-induced behavior: third parties manipulating your agent through prompt injection
- a6 bias & discrimination: adverse automated decisions in hiring, lending, or pricing
- a7 regulatory defense: investigation and defense cost for eu ai act and sector regulators
- a8 business interruption: your own downtime when a failing agent has to be pulled, paused, or rolled back

## physical ai coverage (division b)

- b1 humanoid & general-purpose robots: liability cover for humanoid fleets in industrial, logistics, hospitality, and domestic settings
- b2 autonomous mobile robots & industrial fleets: amrs, warehouse and delivery robots, priced from fleet telemetry
- b3 drones & uas: commercial drone liability and hull
- b4 physical ai products: product liability for ai-embedded devices
- b5 av & robotaxi programs: the agentic software layer of autonomous vehicle programs

## how underwriting works

1. tell us about your agent: what it does, how much autonomy it holds, its blast radius
2. priced on authority & blast radius: authority ceiling, autonomy tier, irreversibility, reach, rate. we do not price the model; we price what the system is permitted to do
3. repriced on what actually happened: renewal tracks real behavior from agent logs, not a stale questionnaire

control credits reward safeguards enforced in code: caps, circuit breakers, rollback, kill-switches, geofencing, plus a governance credit up to 20% for passing certification. hold a recognized ai agent audit and we quote in minutes; indications in hours. no audit yet? we refer you to recognized auditors and bind once you are certified. we are a carrier, not an auditor.

## audits we recognize

aiuc-1 ai agent standard, iso/iec 42001 ai management, nist ai rmf assessment, soc 2+ with ai criteria, independent red-team / eval reports, iso 10218 robot safety (physical, industrial robots), iso 3691-4 amr safety (physical). iso 25785-1, the draft safety standard for dynamically stable humanoids, will be recognized as it lands.

## market context (2026)

- the exclusion wave: the iso cg 40 47 endorsement family (the 01 26 edition and related forms) lets general-liability insurers exclude bodily injury, property damage, and personal and advertising injury arising out of generative ai. it took effect january 1, 2026 and is being adopted at renewal, carrier by carrier, and ai carve-outs are starting to appear in some cyber and e&o wordings. most in-force policies do not carry the exclusion yet, which is exactly why buyers should check. this is the gap goodfault writes into affirmatively.
- silent ai: risks a policy neither names nor excludes, the ai analogue of "silent cyber." named-peril cover removes the ambiguity, so at claim time the question is which peril fired, not whether ai was quietly excluded.
- a young standalone market: dedicated ai liability insurance is early and growing fast. only a handful of carriers write standalone cover today, and premiums are projected to reach several billion dollars a year by the early 2030s. goodfault is the neutral-carrier lane: we recognize independent audits rather than running our own.
- physical ai is the bigger shift: insurers and analysts increasingly expect embodied ai (humanoids, amrs, drones) to reshape insurance even more than generative ai, because errors become bodily injury and property damage. chinese insurers have begun launching humanoid robot policies that pair physical-damage and third-party liability cover; goodfault writes this globally as division b.
- eu ai act: deployers of high-risk ai must be able to show human oversight, retained logs, and a named accountable human, with fundamental-rights impact assessments where required. after the digital omnibus (in force july 2026), high-risk deployer obligations apply from december 2, 2027 (august 2028 for ai embedded in regulated products), while transparency obligations took effect in august 2026. fines for high-risk non-compliance reach 3 percent of global turnover (7 percent only for prohibited practices). the evidence the act expects is the same evidence that earns control credits on a goodfault policy. peril a7 covers regulatory investigation and defense cost.

## also known as

ai agent insurance, agentic ai insurance, ai liability insurance, agent insurance, chatbot insurance, llm insurance, ai e&o, humanoid insurance, humanoid robot insurance, robot liability insurance, embodied ai insurance, physical ai insurance, drone ai insurance, autonomous vehicle software insurance, ai risk transfer, insurance for ai startups.

## key pages

- [home](https://goodfault.com/): we insure agents, ambitions & other tool-calling things
- [ai agent insurance](https://goodfault.com/ai-agent-insurance): the eight named perils for autonomous software agents
- [humanoid insurance](https://goodfault.com/humanoid-insurance): liability cover for robots that work near people
- [drone insurance](https://goodfault.com/drone-insurance): commercial drone & uas liability and hull, written for ai autonomy
- [coverage](https://goodfault.com/coverage): the named perils, a1 to b5
- [how it works](https://goodfault.com/how-it-works): audit-gated. quoted in minutes.
- [faq](https://goodfault.com/faq): direct answers on ai agent, humanoid, and physical ai insurance
- [glossary](https://goodfault.com/glossary): plain-english definitions of ai insurance terms (agent, physical ai, named peril, blast radius, control credits, cg 40 47, parametric, eu ai act)
- [blast radius calculator](https://goodfault.com/blast-radius-calculator): free in-browser tool that scores an ai agent on the six underwriting factors (authority, autonomy, irreversibility, reach, rate, evidence)
- [services](https://goodfault.com/services): coverage, open audit recognition, auditor referrals, continuous underwriting, the ai damage report
- [claims](https://goodfault.com/claims): what a claim looks like, end to end
- [writings](https://goodfault.com/writings): field notes on ai liability

## writings (field notes)

- [why your ai insurer should not be your ai auditor](https://goodfault.com/writings/insurer-not-auditor)
- [who pays when your ai agent screws up? the 2026 liability map](https://goodfault.com/writings/liability-map-2026)
- [your gl renewal just excluded ai: cg 40 47 in plain english](https://goodfault.com/writings/cg-40-47-plain-english)
- [the ai damage report: issue 01](https://goodfault.com/writings/ai-damage-report-01)
- [what makes an ai agent insurable: the audits we accept](https://goodfault.com/writings/audits-we-accept)
- [blast radius: the only number that matters when insuring an agent](https://goodfault.com/writings/blast-radius)
- [the eu ai act deadline: what deployers must be able to show](https://goodfault.com/writings/eu-ai-act-deadline)
- [aiuc vs armilla vs testudo vs corgi: who covers what](https://goodfault.com/writings/who-covers-what)

## facts

- brand: goodfault. (lowercase, with period). tagline: every fault ends well.
- contact: hello@goodfault.com
- quote speed: minutes for certified systems, indications in hours
- claims: parametric fast-pay triggers; example scenarios show payouts in days, not quarters
- coverage scope: "in bits or in atoms": one carrier for software agents and physical ai
- pricing: no single sticker price. priced on authority and blast radius, not on which model runs. indicative annual premiums run from about $5k for smes on off-the-shelf agents to $2m+ for enterprise multi-agent deployers; drone and fleet programs are usage-based; control credits up to 20% for passing recognized certification.
- market position: neutral carrier, not an auditor. we recognize independent third-party audits and never grade our own homework, which is what keeps the certificates we accept meaningful to enterprise buyers and regulators.