PricingAn investment in the product you’ve already built.
Frontier AI security isn’t a feature. It’s the difference between keeping the company you’ve built and explaining to your customers why their data is on a Telegram channel. The average UK data breach now costs £3.4M. We don’t.
£6,000/ month
Or £64,800/yr prepaid · 10% off
- WAF, DDoS, and edge hardening
- Continuous code & dependency scanning
- Bot, scraper, and credential-stuffing defense
- Identity & session hygiene review
- Optional AI / prompt-injection coverage
- Weekly threat & exposure briefings
- Email + WhatsApp support, NBD response
Talk to usMost chosen£18,000/ month
Or £194,400/yr prepaid · 10% off
- Everything in Foundation
- Mythos-class continuous red teaming
- 24/7 monitoring & incident response
- App, API, and infrastructure pentest cycles
- N-day exploit validation against your environment
- Quarterly proof-of-exploit reports
- Dedicated security engineer
- < 4 hour incident response SLA
Start the conversation ↗From £60,000/ month
Custom contract · 12-month minimum
- Everything in Frontier
- Dedicated multi-engineer team
- Private-cloud or on-prem deployment
- Regulatory reporting (FCA, ICO, HIPAA, PCI)
- 1 hour incident response SLA
- Quarterly executive briefings
- Bespoke threat-model development
Request a proposalOne-time engagementsTargeted work, when retained isn’t the answer.
For pre-launch product, M&A diligence, or a single high-stakes surface, we run focused engagements with the same capabilities behind every retained program.
Frontier Audit4 weeks
£35,000
A complete Mythos-class assessment of your live environment. Codebase, dependencies, infrastructure, model boundaries. Delivered as a written report with proof-of-exploit for every finding, ranked by severity and exploitability.
Pre-launch Red Team2 weeks
£25,000
Before you ship. Adversarial assessment of an AI feature, a new product, or a planned launch. Findings within five business days; remediation guidance through to release.
Incident Response RetainerAnnual
£15,000/yr + hourly
On-call when something is on fire. Triage within 60 minutes, forensic capability, regulatory-grade reporting, and a team that has seen the failure mode before.
Why this isn’t cheap
You’re paying for capability that isn’t for sale.
~50Organizations have direct Mythos access
Project Glasswing partners and a small set of vetted critical infrastructure operators. That’s it. Mythos is not on a public pricing page.
£3.4MAverage UK data breach
IBM’s 2024 Cost of a Data Breach Report. Frontier-augmented attacks are not factored into that number yet. They will be.
181 ↑The exploit gap is real
Frontier models produced 181 working browser exploits on a benchmark that prior models hit twice. The defenders who get ahead of this curve don’t pay later.
< 1 dayCVE → working exploit, autonomously
Patch windows that used to last weeks now last hours. We staff and price for that reality, not for the security industry of 2018.
Pricing is indicative for standard engagements and is subject to a scoping call. Mythos and Project Glasswing are research efforts of Anthropic. A.AI accesses Mythos-class capabilities through partners and is not directly affiliated with Anthropic.