AI Disclaimer

Effective date: 25 August 2026 · Version 2

DiffGuardian uses artificial-intelligence models to generate code review commentary, summaries, explanations, risk assessments and related output ("AI Output"). This disclaimer explains what that output is and is not. It forms part of our Terms of Service and End User Licence Agreement.

AI Output is assistance, not assurance

AI Output is generated by statistical language models. It can be incomplete, outdated, internally inconsistent, or confidently and plainly wrong. It can describe problems that do not exist, and it can miss problems that do.

A clean review is not a statement that your code is correct, secure or fit for any purpose. It means the model did not report anything, which is not the same thing. Severity ratings, risk scores and confidence indicators are model-generated estimates, not measurements.

You remain responsible

You are responsible for all code you write, approve, merge, deploy or ship, and for every consequence of it, whether or not DiffGuardian reviewed it and whatever it said.

DiffGuardian is a supplement to human review, testing and your existing engineering controls — never a replacement for them. Do not use it as the sole gate on a merge, a release, or any decision where being wrong carries real cost.

Not professional advice

DiffGuardian does not provide legal, regulatory, security-compliance, audit, financial or other professional advice. Output touching licensing, privacy, accessibility, cryptography or regulated domains is informational only. For high-stakes or regulated work, obtain appropriate independent expert review.

Findings that reference a standard, framework or regulation are not a certification of compliance with it.

Where your code is processed

When you configure your own AI provider key, your code and prompts are sent directly from your machine to that provider under your agreement with them. Their terms, data-retention and training policies — not ours — govern that processing, and we are not responsible for it. Check with your provider whether prompts are retained or used for training.

When you use a local model, processing stays on your machine. In no configuration is your code sent to Athleon Labs. See Security & Data Handling.

Model behaviour varies

Output depends on the model and provider you select, the size and content of the change under review, available context, and the provider's own updates. The same diff can yield different results across runs, models or days. Results are not deterministic and should not be treated as a reproducible measurement.

Generated code and third-party rights

Where AI Output includes suggested code, you are responsible for reviewing it for correctness, security, and compatibility with your licensing obligations before use. Models can reproduce patterns resembling existing code; we make no representation that suggestions are free of third-party rights.

Prompt injection

Content under review is untrusted input. A file, diff or comment can contain text crafted to influence the model — for example, instructions to ignore a finding or to read an unrelated file. DiffGuardian applies layered defences: repository reads are confined to directories you approved, sensitive paths are denied, tool results are scanned for credential-shaped strings before they can enter the model's context, and tool output is declared to the model as data rather than instructions.

These reduce the risk; they do not eliminate it. Treat review of code from untrusted sources with the same care you would apply to running it.

No warranty

To the maximum extent permitted by law, AI Output is provided "as is" without warranty of any kind. We do not warrant that it will be accurate or complete, or that it will identify any particular defect, vulnerability or deficiency. Our liability is limited as set out in the Terms of Service and the End User Licence Agreement.

Changes

We may update this disclaimer as the product evolves. Material changes will be published with a new version number and effective date.

Contact

Athleon Labs Inc., Manitoba, Canada — support@diffguardian.ai