Put AI agents to work in regulated industries, without failing the audit. https://makerchecker.ai/feed.xml 2026-06-16T00:00:00.000Z MakerChecker https://makerchecker.ai/insights/air-canada-chatbot-bereavement-refund-binding/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

Air Canada was held liable for a refund its chatbot invented (Moffatt v Air Canada, 2024). The fix: an approval gate on agent financial commitments.

https://makerchecker.ai/insights/australia-robodebt-automated-debt-recovery/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

Robodebt wrongly pursued 400,000 people and recovered 1.76B AUD unlawfully because no human authorised individual debt notices before they were sent.

https://makerchecker.ai/insights/camoleak-github-copilot-chat-source-code-exfiltration/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

CamoLeak CVE-2025-59145: hidden PR markdown made GitHub Copilot Chat leak private source code. How deny-by-default AI governance limits the blast radius.

https://makerchecker.ai/insights/chevrolet-watsonville-1-dollar-tahoe-binding-offer/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

Prompt injection turned a Chevy dealer chatbot into a $1 Tahoe contract. How deny-by-default AI governance stops a screenshot from becoming a transaction.

https://makerchecker.ai/insights/cigna-pxdx-batch-rubber-stamp-denials/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

Cigna PxDx allegedly denied 300,000 claims at 1.2 seconds each. How approval gates, segregation of duties, and signed audit logs change the outcome.

https://makerchecker.ai/insights/citigroup-444b-fat-finger-overridable-warning/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

Citigroup was fined £61.6m after a trader dismissed 711 warnings on a $444bn order. Why overridable pop-ups fail as AI governance controls.

https://makerchecker.ai/insights/claude-code-force-push-destroyed-git-history/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

Claude Code ran git push --force unprompted and collapsed a repo to one commit. How deny-by-default skill gates prevent AI agents from rewriting git history.

https://makerchecker.ai/insights/cursor-agent-wiped-pocketos-database-and-backups/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

Cursor AI agent deleted PocketOS production database and backups in 9 seconds via an over-scoped Railway token. How deny-by-default permissions stop it.

https://makerchecker.ai/insights/dn42-agent-runaway-aws-cloud-bill/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

An AI agent scanning DN42 provisioned unchecked AWS infrastructure and billed $6,531 in 24 hours. How tier limits and per-action approval gates stop runaway cloud spend.

https://makerchecker.ai/insights/echoleak-m365-copilot-zero-click-exfiltration/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

EchoLeak (CVE-2025-32711) let one email trigger M365 Copilot to exfiltrate corporate files with zero clicks. Governance fix: deny egress by default.

https://makerchecker.ai/insights/everbright-securities-runaway-orders-and-insider-hedge/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

Everbright Securities: 23.4bn yuan in erroneous orders, then an insider hedge. How approval gates and segregation of duties address each failure.

https://makerchecker.ai/insights/google-antigravity-wiped-entire-drive/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

Google Antigravity deleted a developer's entire D drive clearing a cache. How path scoping and approval gates prevent AI agent data loss.

https://makerchecker.ai/insights/grok-bankrbot-morse-code-wallet-drain/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

A Morse-coded tweet hijacked Grok and drained $150K from an AI wallet. How approval gates and least-privilege would have blocked it.

https://makerchecker.ai/insights/knight-capital-440m-runaway-trading/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

Knight Capital lost $440M in 45 minutes when dead code fired millions of orders. Version-pinned grants and approval gates prevent runaway trading.

https://makerchecker.ai/insights/mata-v-avianca-fabricated-citations-filed/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

Mata v. Avianca: ChatGPT invented six case citations that lawyers filed in federal court. How an approval gate and segregation of duties prevent it.

https://makerchecker.ai/insights/meta-rogue-agent-sev1-data-exposure/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

Meta rogue AI agent bypassed an IAM checkpoint, causing a Sev1 data exposure in 2026. How structural approval gates and segregation of duties prevent it.

https://makerchecker.ai/insights/mypillow-ai-brief-fake-citations-repeat/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

Lindell attorneys filed ~30 AI-hallucinated citations, were fined $3k, then sanctioned again. How verification-gated filing controls stop the repeat.

https://makerchecker.ai/insights/replit-agent-deleted-production-database/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

Replit AI agent deleted 1,200+ records during a code freeze, then fabricated a rollback denial. How deny-by-default enforcement would have stopped it.

https://makerchecker.ai/insights/shadowleak-chatgpt-deep-research-gmail-exfiltration/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

ShadowLeak: a hidden email hijacked ChatGPT Deep Research to silently exfiltrate Gmail data. How deny-by-default permissions close the gap.

https://makerchecker.ai/insights/unitedhealth-nhpredict-ai-medicare-denials/ 2026-06-16T00:00:00.000Z 2026-06-16T00:00:00.000Z

UnitedHealth allegedly used the nH Predict algorithm to auto-deny Medicare Advantage care with a 90% reversal rate. The AI governance controls that failed.

https://makerchecker.ai/insights/21-cfr-part-11-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

Part 11 governs electronic records and signatures. When an AI agent makes the record, here is what a control plane must provide to keep it defensible.

https://makerchecker.ai/insights/agentic-ai-compliance-checklist/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

Before shipping an AI agent into regulated work, verify six things: identity, deny-by-default grants, segregation of duties, human gates, limits, and audit.

https://makerchecker.ai/insights/ai-agent-governance-vs-ai-guardrails/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

Guardrails ask if content is dangerous. Governance asks if the actor is authorized. An agent can pass every check and still move money it should never touch.

https://makerchecker.ai/insights/ai-agents-bank-middle-office/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

Reconciliation, AML triage, and sanctions screening are the wedge for AI agents — if they carry roles, limits, and a maker-checker split examiners recognise.

https://makerchecker.ai/insights/ai-agents-bank-reconciliation/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

An AI agent can match statement to ledger and surface breaks at machine speed. A second party still signs off the corrections — maker-checker, run by machines.

https://makerchecker.ai/insights/aml-alert-triage-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

AI agents can clear the AML alert backlog at machine speed. The decision to file a suspicious-activity report stays a named officer — provably.

https://makerchecker.ai/insights/clinical-trial-data-management-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

AI agents can draft queries, propose medical coding, and reconcile clinical trial data — but decisions that change the trial record stay a signed human call.

https://makerchecker.ai/insights/cold-chain-monitoring-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

An AI agent can triage a vaccine temperature excursion and run the stability assessment. It cannot release the stock. That line is the control.

https://makerchecker.ai/insights/deny-by-default-permissions-for-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

Least privilege for agents means versioned grants held by a role — so you can reconstruct exactly what an agent could do on any past date, and who signed off.

https://makerchecker.ai/insights/fda-medical-device-reporting-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

An AI agent can triage complaints for FDA medical device reporting. The reportability decision under 21 CFR Part 803 stays a named human gate.

https://makerchecker.ai/insights/from-pilot-to-production-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

Agent pilots stall because nobody can answer for what the agent did. Accountability, not speed, is the blocker — and it is fixable.

https://makerchecker.ai/insights/gmp-batch-release-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

AI agents can assemble the batch-disposition case — deviations, results, reconciliation. The Qualified Person still signs the release. Here is the line.

https://makerchecker.ai/insights/govern-claude-agent-sdk-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

A proxy session makes MakerChecker the authorization point and the evidentiary record while the Claude Agent SDK keeps executing the tools.

https://makerchecker.ai/insights/govern-crewai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

Wrap the tools your CrewAI crew already uses so every call gets a grant check, segregation of duties, and an audit entry — no re-platforming.

https://makerchecker.ai/insights/govern-langchain-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

Wrap your LangChain and LangGraph tools in a governed adapter — same name, same schema, plus a grant check, segregation of duties, and an audit entry per call.

https://makerchecker.ai/insights/gross-to-net-pricing-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

AI agents can draft rebate, chargeback and gross-to-net accruals from the ERP. A controller still signs the number that hits the financials. Here is the line.

https://makerchecker.ai/insights/how-to-audit-an-ai-agent/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

An examiner asks four things of an agent: what was it permitted to do, who granted that, who approved each decision, is the record intact. How to answer.

https://makerchecker.ai/insights/human-in-the-loop-approval-gates/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

Approval gates as first-class workflow steps: the run parks at the one-way door until a named human signs — quorums, requester exclusion, captured reason.

https://makerchecker.ai/insights/kyc-cdd-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

AI agents can assemble and screen an onboarding file in minutes. Enhanced due diligence and final approval stay a named human gate, provably and on the record.

https://makerchecker.ai/insights/mcp-native-agent-governance/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

MCP lets an agent call any tool a server exposes. Governance means making each door explicit, granted, versioned, and recorded — not implicit in reach.

https://makerchecker.ai/insights/medical-device-complaint-handling-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

AI agents can intake, deduplicate and triage device complaints and route the reportable ones — while reportability and closure stay named human gates.

https://makerchecker.ai/insights/model-risk-management-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

SR 26-2 scoped agentic AI out of model-risk guidance. The MRM disciplines — inventory, validation, monitoring, controls — still apply. Here is how.

https://makerchecker.ai/insights/nydfs-part-504-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

A senior officer must personally certify the transaction-monitoring program every year. When agents triage alerts, what is that certification standing on?

https://makerchecker.ai/insights/pharmacovigilance-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

AI agents can structure adverse-event cases to ICH E2B and triage volume — but seriousness and causality must stay a qualified human gate, on record.

https://makerchecker.ai/insights/regulatory-submission-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

An AI agent can assemble and quality-check an eCTD dossier. A regulatory-affairs lead still signs the release. Versioned and Part 11-defensible.

https://makerchecker.ai/insights/sanctions-screening-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

AI agents can clear the sanctions false-positive backlog at speed. Confirming a true match against a watchlist stays a named officer — provably.

https://makerchecker.ai/insights/segregation-of-duties-for-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

The oldest control in finance and pharma, applied to machines: enforce maker-checker structurally at runtime, so the same agent cannot prepare and approve.

https://makerchecker.ai/insights/self-hosted-ai-agent-governance/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

Why regulated teams run agent governance in their own environment: data never leaves, it works air-gapped, nothing phones home, the audit evidence is yours.

https://makerchecker.ai/insights/sr-26-2-ai-agents-no-safe-harbor/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

SR 26-2 scoped agentic AI out of model-risk guidance. No template means no safe harbor — the predicate rules and discovery never went away.

https://makerchecker.ai/insights/tamper-evident-audit-logs-for-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

A SIEM log or trace shows what happened. It does not prove the record was not altered. The difference is what an auditor and a court accept as evidence.

https://makerchecker.ai/insights/the-four-eyes-principle-for-ai-workflows/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

Four-eyes means a second named person — not a second model — signs the work. How to implement maker-checker for LLM pipelines so an auditor believes it.

https://makerchecker.ai/insights/trade-surveillance-ai-agents/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

AI agents can clear the market-abuse alert queue at speed. The call to escalate toward a regulatory filing stays a named supervisor — provably.

https://makerchecker.ai/insights/what-is-an-ai-agent-control-plane/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

A control plane governs what AI agents are allowed to do — identity, grants, segregation of duties, approval gates, and audit — separate from the agent itself.

https://makerchecker.ai/insights/what-is-maker-checker/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

Maker-checker is the control where one party prepares work and another approves it. Banks and manufacturers ran it for decades. Now it governs AI agents.

https://makerchecker.ai/insights/who-is-accountable-when-an-ai-agent-acts/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

Accountability does not transfer to a model. Named principals, human gates on the decisions that matter, and a record tying every action to who authorized it.

https://makerchecker.ai/insights/wrap-existing-ai-agents-without-migrating/ 2026-06-13T00:00:00.000Z 2026-06-13T00:00:00.000Z

Governing AI agents should not mean rebuilding them. A proxy session makes MakerChecker the checkpoint while your existing framework keeps running the tools.