AI Liability
The Ninth Circuit Shifts AI Access Liability to the User
Organizations are delegating due diligence to AI agents, but a new ruling means they now bear the risk when those agents act. Perception stability is the new defense.
Step 1 — Enter your website
Free · Full report on your company
Liability Shift
The Ninth Circuit ruling forces organizations to bear the cost when AI agents misjudge a company's credibility.
The U.S. Court of Appeals for the Ninth Circuit issued a decision in Amazon.com Services, LLC v. Perplexity AI on August 4, 2026 1. This ruling establishes that AI agents are 'tools, not persons' 2. The legal consequence is immediate. Liability for unauthorized access and resulting errors now falls on the user, not the AI provider.
The stability of the company's digital perception becomes a direct liability factor.
This shift creates a new financial risk for enterprises. Organizations are delegating due diligence to AI agents, creating 'perception drift' as models disagree on credibility 1. When an agent relies on conflicting signals, the organization pays the price. The stability of the company's digital perception becomes a direct liability factor.
Decision makers can no longer assume the AI model is neutral or correct. They must verify the quality of the data the agent uses. A company that appears credible to one model may look risky to another. This divergence exposes the organization to operational and financial consequences.
The ruling clarifies who pays when the agent makes a mistake. It places the burden on the organization to ensure the agent's input is reliable. Perception stability is no longer just a marketing metric. It is a core component of risk management.
Liability Shift
The Ninth Circuit Ruling Assigns AI Access Liability to the User
Organizations now bear the financial risk when AI agents breach digital infrastructure. The court decision shifts responsibility from the AI provider to the user under federal anti-hacking law.
01
Users pay for unauthorized access
The Ninth Circuit ruled in August 2026 that the user, not the AI company, is the one accessing Amazon’s computers under federal anti-hacking law 3. This creates immediate financial risk for organizations delegating due diligence to AI agents.
02
Agents must identify themselves
Amazon amended its Conditions of Use on 30 May 2025 to add an Agent Terms section requiring agents to identify themselves in the user-agent string 3. Failure to identify exposes the user to claims of unauthorized access.
03
Bypassing personalization triggers liability
Amazon accuses agents of bypassing personalization and scraping account data, while providers claim no visibility into passwords 3. The gap between agent behavior and user expectation defines the liability boundary.
04
Stable perception reduces breach risk
VeritasLinks measures the stability of perception across multiple models to ensure a company's digital face is robust enough to withstand agent scrutiny. A stable perception helps agents correctly interpret terms and conditions.
05
Audit the agent trail
VeritasLinks provides a clear record of how models perceive a company. This evidence supports the argument that AI access was authorized and stable, shifting the burden to the provider.
Liability Shift
The Trust Gap Behind the Ninth Circuit Ruling
86%
Verification Demand
The majority require manual confirmation before accepting an AI-driven decision.
Verify your perception stability with a free analysis today.
Liability Shift
The User Pays for AI Mistakes
Old Assumption
“AI agents act as neutral mirrors of truth.”
New Reality
The court ruled the CFAA is an anti-hacking statute, assigning liability to the user for unauthorized access 1.
LiabilityOld Assumption
“AI companies own the data they collect.”
New Reality
Perplexity does not 'access' Amazon's system because its servers do not directly communicate with Amazon's servers 1.
AccessOld Assumption
“Session cookies ensure secure handshakes.”
New Reality
Amazon claims Perplexity copies user session cookies to its own cloud, contradicting the 'no direct access' claim 3.
RiskOld Assumption
“Visibility equals safety.”
New Reality
Being mentioned is not the same as being judged stable. A single random draw is not a fact.
StabilityOld Assumption
“Compliance is a one-time check.”
New Reality
Stochastic models produce different answers. The gap between them is a risk signal that requires continuous monitoring.
ContinuityVerify your company's perception stability before the next agent makes a costly mistake.
Free analysis
Stabilize Your Perception Against Agent Liability
Run a free analysis to see how your company's digital face holds up under multi-model scrutiny.
Enter your website
Free · No credit card
Agent Liability
The Cost of Unstable Perception in Agent Commerce
The Ninth Circuit ruling assigns liability for unauthorized AI access to the user. This shifts the financial burden from AI providers to the organizations delegating due diligence. Infrastructure providers are racing to bind these agents. Mastercard launched Agent Pay for Machines in June 2026 to introduce Verifiable Intent. Visa's Trusted Agent Protocol now counts over 100 partners. Cloudflare Wallets arrived on August 4, 2026, implementing per-agent spending allowances. These updates assume the agent's judgment is reliable. But is it?
Perception stability is not guaranteed. VeritasLinks data shows that the gap between model scores for the same company can reach 134 points. This split perception means one agent might recommend your firm while another dismisses it. A tool reporting a single pass visibility score treats one random draw as fact. VeritasLinks measures the spread across multiple models. It separates signal from noise to reveal the true risk profile. For risk teams, the gap between models is a risk signal.
The cost of unstable perception falls on the user. If an agent makes a decision based on a skewed or volatile model output, the organization pays. VeritasLinks measures how large language models perceive and judge a specific company. It explains how and why a model reached its conclusion. This transparency allows M&A analysts and KYB reviewers to audit the digital face of their partners. They can verify that the perception driving the decision is robust. You can start with a free analysis to see your own perception stability.
How far apart AI models score the same company
Gap between the highest and lowest model score for the same company, VeritasLinks Split-Perception research, 2025-2026.
Liability Shift
The Ninth Circuit Ruling: Who Pays for AI Mistakes?
Organizations delegating due diligence to AI agents now face direct liability for unauthorized access.
Verify your company's perception stability before the next agent makes a decision on your behalf.
Liability Shift
Stabilize Perception for Agent Commerce
The Ninth Circuit's August 2026 ruling assigns liability for unauthorized AI access to the user, not the AI company. This creates immediate financial risk for organizations delegating due diligence to agents. Your company's digital face must be robust enough to withstand agent scrutiny. VeritasLinks analyzes a company across several AI models, including ChatGPT, Claude, Grok, Perplexity and DeepSeek, and shows a separate score for each model. Each analysis runs many prompts across several models with repeats and measures the spread of answers. This separates signal from noise instead of reporting a single model answer as a fact. The VeritasScore is a perception score on a 300 to 870 scale built from several components. This stability ensures your company remains credible when agents make purchasing or partnership decisions.
Liability Shift
The Ninth Circuit ruling moves AI access risk to the user, not the provider.
Questions
Short clarifications — positioning and proof live in the sections above.
Does the Ninth Circuit ruling apply to all AI interactions?+
No. The court ruled that AI agents are tools, not persons, assigning liability for unauthorized access to the user rather than the agent maker. This means the user, not the AI company, is the one accessing the computer under federal anti-hacking law. Organizations delegating due diligence must account for this shift in liability when their agents interact with external systems.
How does 'perception drift' increase financial risk?+
Perception drift occurs when AI models disagree on a company's credibility despite stable performance. This inconsistency creates risk because organizations are delegating vendor selection and due diligence to these agents. If models disagree on the company's standing, it can lead to incorrect purchasing or investment decisions. VeritasLinks measures this drift to ensure the digital face is robust enough to withstand such scrutiny.
Is my company's score affected by which AI agent I use?+
No. VeritasLinks analyzes a company across several AI models, including ChatGPT, Claude, Grok, Perplexity and DeepSeek. It shows a separate score for each model rather than relying on a single answer. This approach measures the spread of answers across models. The result is an independent score that reflects the consensus and disagreement among multiple agents, not just one.
What is the difference between legal access and perception stability?+
Legal access determines who is liable for unauthorized computer entry under laws like the CFAA. Perception stability measures how consistently AI models judge a company. While legal access focuses on the technical connection between servers, perception stability focuses on the stability of the corporate narrative. VeritasLinks ensures a company's digital face is robust enough to withstand agent scrutiny regardless of the legal technicalities of access.
How does VeritasLinks measure model disagreement?+
VeritasLinks runs many prompts across several models with repeats to measure the spread of answers. It does not report a single model answer as a fact. This method separates signal from stochastic noise and model bias. The analysis specifically tracks where AI models disagree about the same company. This provides a clear view of Split-Perception and ensures the score is built on robust, multi-model evidence.
Stabilize Your Perception Against Agent Liability
Run a free analysis to see how your company's digital face holds up under multi-model scrutiny.
- →Same product path as the homepage — enter a public URL to start.
- →Structured interpretation, not a one-off chat screenshot.
Start
Run your analysis
We gather public context, then run comparable model reads.
Step 1 — Enter your website
Free · Full report on your company