Tim McAllister · AI trust in practice
AI Trust
Market Report.
The only constant is change. Get the report and follow the updates as the AI security and trust market evolves.
AI Trust Market Report
Get the report and future updates.
Confirm your email to download the PDF. I’ll send a short email when an updated report is ready, with a link and what changed. Free. Personal or work email is welcome.
No website account is needed. Kit provides the PDF after email confirmation. Unsubscribe from report updates at any time.
Kit handles email confirmation and unsubscribe links. Each topic is a separate opt-in. Read the privacy notice.
Unsubscribing stops future report updates; it does not remove copies already downloaded.
The publication record
Release history
AI Security and Trust Market
Current versionv1.0.0
An enterprise control map, viewed through a DigiCert AI Trust lens
By Tim McAllister. Report date: . Research cutoff: .
A public-source analysis of enterprise AI security and trust, focused on where DigiCert’s documented agent identity, runtime policy, signing and provenance capabilities fit. It maps nineteen architectural categories, compares complementary control roles, examines dated market changes and classified incidents, and proposes measurable buyer tests. Author analysis is separated from vendor claims and documented facts.
Scope, method, findings and sources
Scope
Enterprise AI agents, their data and execution paths, software/model integrity and content provenance. Principal market-news window July 20–October 18, 2026; historical context explicitly dated. Nineteen author-defined categories; no exhaustive census or market-share estimate.
Method
Selective desk research using public issuer filings, provider/project documentation, corporate announcements, advisories and first-party incident/research accounts accessed October 18, 2026. Documented facts, vendor claims, research demonstrations and author analysis are distinguished. No product efficacy tests, customer interviews or confidential inputs underpin public claims.
Key findings
- Tim’s editorial analysis
Buy the complete control path: authenticated identity, scoped policy, an enforcement point at the resource, measurable containment and retained evidence.
Sources: AI Trust Manager: September 15 release notes, Getting started with Policy in AgentCore, NVIDIA OpenShell
- Vendor claim
DigiCert AI Trust Manager has documented agent discovery, passports, policies and runtime enforcement; the current site still labels Model Trust as preview.
Sources: AI Trust Manager: September 15 release notes, AI Trust Manager product page: Model Trust preview
- Documented fact
Data and non-human identity are converging, but transaction status matters: Proofpoint/Varonis is pending; Cyera/Oasis and Keyfactor/Cofide are closed.
Sources: Varonis Form 8-K: merger agreement with Proofpoint, Cyera completes acquisition of Oasis Security, Beyond static secrets: Keyfactor closes Cofide acquisition
- Tim’s editorial analysis
Identity and signatures leave business intent, data entitlements, semantic inspection, execution isolation and recovery as explicit integration and evaluation questions.
Sources: AI Agent Trust: identity, security and governance, Cisco AI Defense data sheet, NVIDIA OpenShell, OpenSSF Model Signing v1.0 launch
Capabilities covered
- AI discovery and posture
- Testing, red teaming and evaluation
- Detection and response for AI
- Deepfake and synthetic-identity defense
- Agent identity, NHI and secrets
- Action authorization and policy
- AI governance, risk and compliance
- SaaS and agent-platform native controls
- Runtime guardrails and AI/MCP gateways
- Data security for AI
- Workforce AI and browser controls
- Execution isolation and sandboxes
- DNS and network trust
- OT, device and cyber-physical trust
- Software, model and AI supply-chain integrity
- PKI, certificate lifecycle and crypto-agility
- Content provenance and authenticity
- Confidential computing and attestation
- Cross-organization agent trust and commerce
References
- AI Trust Manager general-availability announcement
Published . Accessed .
- AI Trust Manager: September 15 release notes
Publication date unavailable. Accessed .
- AI Agent Trust: identity, security and governance
Publication date unavailable. Accessed .
- AI Trust Manager product page: Model Trust preview
Publication date unavailable. Accessed .
- DigiCert Brings AI Passports to the NVIDIA Open Agent Safety Platform
Published . Accessed .
- Software Trust Manager
Publication date unavailable. Accessed .
- Content Trust Manager
Publication date unavailable. Accessed .
- Independent trust validation with Google Cloud
Published . Accessed .
- Varonis Form 8-K: merger agreement with Proofpoint
Published . Accessed .
- Proofpoint to acquire Varonis: joint release, Exhibit 99.1
Published . Accessed .
- Cyera completes acquisition of Oasis Security
Published . Accessed .
- Bring data and identity together to secure the agentic enterprise
Published . Accessed .
- Beyond static secrets: Keyfactor closes Cofide acquisition
Published . Accessed .
- NVIDIA launches Open Agent Safety Platform
Published . Accessed .
- NVIDIA OpenShell
Publication date unavailable. Accessed .
- Getting started with Policy in AgentCore
Publication date unavailable. Accessed .
- Stop a running AgentCore session
Publication date unavailable. Accessed .
- What is Microsoft Entra Agent ID?
Publication date unavailable. Accessed .
- Cisco AI Defense data sheet
Publication date unavailable. Accessed .
- Prisma AIRS documentation
Publication date unavailable. Accessed .
- SPIFFE overview
Publication date unavailable. Accessed .
- OpenSSF Model Signing v1.0 launch
Published . Accessed .
- RFC 9334: RATS Architecture
Publication date unavailable. Accessed .
- C2PA Technical Specification 2.4
Publication date unavailable. Accessed .
- NIST AI Risk Management Framework
Publication date unavailable. Accessed .
- Compromised PyTorch-nightly dependency chain
Published . Accessed .
- GitHub MCP exploited: accessing private repositories via MCP
Published . Accessed .
- AWS-2025-015: Amazon Q Developer extension
Published . Accessed .
- Data theft via Salesloft Drift
Published . Accessed .
- Opsin announces $21M funding
Published . Accessed .
- Harmonic announces $50M Series B
Published . Accessed .
- Pluto emerges from stealth with $42M total funding
Published . Accessed .
- Rein announces Series A
Published . Accessed .
- Rein company-issued Series A announcement
Published . Accessed .
- Kontext announces $4M financing
Published . Accessed .
- Outerlimit emerges from stealth with pre-seed funding
Published . Accessed .
- Eve extends seed round to $7.5M
Published . Accessed .
- AIUC announces Series A
Published . Accessed .
- Exein announces financing for physical-AI security
Published . Accessed .
- Zenity announces $125M Series C
Published . Accessed .
- Okta Agent Gateway runtime security
Published . Accessed .
- Aembit extends access controls to personal AI agents
Published . Accessed .
- Zenity Labs discloses AgentCorruption
Published . Accessed .
- Mistral security advisory MAI-2026-003
Published . Accessed .
- DIVD case DIVD-2026-00014
Published . Accessed .
- Zammad advisory: CVE-2026-102489 / CVE-2026-102490
Published . Accessed .
- Hugging Face July security incident
Published . Accessed .
- OpenAI acknowledgment of Hugging Face model-evaluation incident
Published . Accessed .
- Hugging Face incident and the road ahead
Published . Accessed .
- Shared clipboard inside the sandbox: cross-account leakage research
Published . Accessed .
- APSB26-147: Content Credentials SDK security update
Published . Accessed .
What changed
- First personally authored public-source edition with DigiCert-focused analysis.
- Rebuilt public evidence, taxonomy, availability distinctions and buyer tests; no confidential source package redistributed.
Cite this version: Tim McAllister. AI Security and Trust Market. v1.0.0. Published October 18, 2026 at 09:15:11 PM UTC. Public version citation.
Confirm your email to download the PDF →- Tim’s editorial analysis
What to expect
One reference for each release
Each edition keeps a public version reference on this page. New findings and corrections appear in a new version, with a change summary. Confirm your email to download the current PDF.
A separate subscription
Report updates and AI incident alerts are separate choices. Signing up for one does not subscribe you to the other.
Sources and judgment
Each release will explain its scope and method, link to dated references, and distinguish documented facts, vendor claims, and my analysis.
Clear authorship
Each release will state its affiliations and disclosures. My independent analysis will be identified separately from any guidance approved by my employer.