Disclaimer
Agent Safety Score is run by BitssyBytes. This page explains what our grades are, what they aren't, and how to tell us when we got something wrong. Using the site is also covered by our Terms of Use and Privacy Policy.
Grades are our opinion, applied to dated public sources
Every grade, score, label, and note on this site and in our newsletter is our editorial opinion. We form it by reading the public pages listed on each tool page and applying the published rules on our methodology page. Each tool page shows the date we last checked it and links every source we used. A grade describes what those sources said on that date. It is not a statement about anything those sources don't cover.
Where we report a fact, such as what a vendor's documentation says, a vendor's own statement about an incident, or what a news outlet reported, we attribute it and link it so you can check it yourself.
What a grade is not
A grade is not a security audit, penetration test, certification, or guarantee. We don't test products. We describe what a vendor's documentation says you can configure, not how well it works in practice. "Unverified" means we couldn't confirm a control in public docs. It does not mean the control is missing.
Not professional advice
Nothing here is security, legal, compliance, financial, or any other professional advice. Before you choose, deploy, or buy a tool, test it yourself and talk to someone qualified for your situation.
Grades can be out of date or wrong
AI tools change fast, and documentation can be incomplete, unclear, or changed without notice. A grade may no longer match the product, and we can simply get things wrong. Check the linked sources and the last-checked date before you rely on anything here.
Corrections and vendor responses
If you think a grade, note, or label is wrong or out of date, tell us. Our Corrections & Disputes policy explains how to send a correction, how we review it, and how a vendor can add a short response. You can also use the submit form or email [email protected]. We review every correction request. Sponsors use the same process as everyone else.
No affiliation or endorsement
We are not affiliated with, endorsed by, or paid by any vendor we list, except through clearly labeled sponsorship, which never affects grades. Product names and trademarks belong to their owners. We use them only to identify the products we write about. A listing, good or bad, isn't an endorsement or criticism of a company beyond what its scorecard says.
Vendors may not use our name, logo, or grades in advertising in a way that suggests we endorse them. Linking to your tool page is fine.
How we use AI tools
We use AI tools to help research and draft, including Grok Bot, which is also a product we rate. Each published update is checked against the cited public sources by a separate review pass before it goes live. If we have any business relationship with a vendor we rate beyond sponsorship, we disclose it on that tool's page.
Sponsorship is labeled and kept separate
We sell one sponsor spot on the site and in the Friday newsletter. Every sponsored placement says "Sponsored", sits apart from grades, and links with rel="sponsored nofollow". Sponsors don't see grades early, can't change a grade, note, or review date, and don't get faster reviews. A vendor we grade can sponsor, and its tool page then carries a permanent line saying when. See the Sponsor Terms and the sponsor rules.
Links to other sites
We link to vendor docs, news, and sponsor sites so you can check our sources. We don't control those sites and aren't responsible for their content, privacy practices, or availability. Sponsor links are paid placements and are labeled.
No warranty and limits on liability
The site and newsletter are provided "as is" and "as available". To the extent the law allows, we disclaim all warranties, and BitssyBytes's liability is limited as set out in our Terms of Use. The full terms are in our Terms of Use. Some places don't allow these limits, so they apply only as far as the law where you live permits.