- Home
- LLM Security Tools
- GLACIS
GLACIS
What it is
GLACIS adds runtime controls to AI actions and creates signed records of covered decisions for external verification. It can let actions proceed unchanged or, when configured bounds are crossed, constrain them, pause them for a person, or deny them. The service says it runs nine production components while working with a customer’s existing identity, cloud, and security systems. Records use OVERT, a published open format, and can be checked in a browser without a GLACIS account. The site says payloads remain in the customer’s environment and records contain hashes; omitting protected fields depends on deployment setup. The product is offered through API, web, and self-hosted options, with a hybrid deployment model. Its free account is free, with no card required, and it offers a 14-day trial. GLACIS serves AI builders and operators, AI sellers to regulated buyers, healthcare AI overseers, and product evaluators.
For and against
For
- Can constrain, hold, deny, or pass actions
- Signed records use the open OVERT format
- Browser verification does not require a GLACIS account
- Free account at no cost
- API, web, and self-hosted options
Against
- No automated evidence-pack export
- Record checks do not establish safety or compliance
- Protected-field omission depends on deployment setup
- No bug bounty
The verdict
GLACIS combines runtime controls with signed decision records and supports API, web, and self-hosted deployment. Its verification records have stated limits, and evidence packs must be assembled by hand.
Who it suits
GLACIS may suit AI-agent builders or operators and teams selling AI to healthcare or other regulated buyers. It also lists healthcare AI overseers and people evaluating the product among its intended audiences.
Plans and prices
Free account
No card required · setup about a minute
Read 4 Oct 2026 · glacis.io
On the record
- Control decisions
- For actions that cross configured bounds, Glacis can constrain, hold for a person, or deny the action; it can also allow actions through unchanged.glacis.io · read 4 Oct 2026
- Data handling
- The site says payloads stay in the customer’s environment and records carry hashes; protected fields can be left out, depending on deployment setup.glacis.io · read 4 Oct 2026
- Deployment evaluation
- The contact page describes a six-week evaluation on the workflow a customer is reviewing.glacis.io · read 4 Oct 2026
- Evidence pack limitation
- Glacis does not offer automated evidence-pack export today; packs are assembled by hand for a particular review.glacis.io · read 4 Oct 2026
- Managed service
- Glacis says it runs nine production components as a managed service, while working with the identity provider, cloud, and security systems the customer already uses.glacis.io · read 4 Oct 2026
- Open standard
- OVERT is a published open record format that implementers and reviewers can inspect and use without a proprietary schema.glacis.io · read 4 Oct 2026
- Records
- Records are signed and use OVERT, an open record format; a verifier can check a record in a browser without a Glacis account.glacis.io · read 4 Oct 2026
- Security reporting
- The vulnerability disclosure policy says Glacis aims to acknowledge reports within three business days and provide an initial assessment within ten business days; it offers no bug bounty.glacis.io · read 4 Oct 2026
- Support
- Glacis says it typically replies to contact requests within one business day and takes on a small number of managed deployments each quarter.glacis.io · read 4 Oct 2026
- Verification limits
- A record check does not establish whether an action was correct, every event was captured, or a system is safe or compliant.glacis.io · read 4 Oct 2026
- What it does
- Glacis places runtime controls in the path of AI actions and creates signed records of covered decisions for external verification.glacis.io · read 4 Oct 2026
- Who it serves
- The signup form lists AI sellers serving healthcare or other regulated buyers, healthcare AI overseers, AI-agent builders or operators, and people evaluating the product.glacis.io · read 4 Oct 2026
