NeuralTrust has raised $20M to secure the growing swarm of AI agents in the enterprise. We're building the security layer that the world's most critical enterprises will run on. Is your security ready for the agentic era?

Switching from ChatGPT to Claude? Your memory doesn’t move automatically. Follow these five steps: 1. Review what ChatGPT saved 2. Ask it to write it out in one code block 3. Remove anything sensitive first 4. Settings > Your Claude > Start import 5. Check what Claude actually learned Note: Anthropic calls this experimental. The ChatGPT .zip export doesn’t include memories. Full guide → neuraltrust.ai/blog/move-ai-… #ChatGPT #Claude #AIMemory

When agents do the work, where does the know-how live? In the vendor’s memory feature. Tied to someone’s account. 30-day version window. No shared export format. That’s the institutional memory problem. Not that agents forget, it’s that what they learn ends up somewhere you don’t control. Full breakdown → neuraltrust.ai/blog/institut… #AgenticAI #InstitutionalMemory #EnterpriseAI

NeuralTrust is positioned in the Catalysts quadrant of the ITMAP Agentic AI Security map. Second Catalyst recognition from ITMAP this month: AI Gateways last week, Agentic AI Security today. Catalysts = strong Product Strategy and Execution. 15 vendors on the map. 👉 neuraltrust.ai/news/neuraltr… #AgenticAI #AIRuntimeDefense #AISecurity #CISO

𝗔𝗻 𝗮𝗴𝗲𝗻𝘁 𝗶𝘀 𝗮 𝗺𝗼𝗱𝗲𝗹 𝗽𝗹𝘂𝘀 𝗮 𝗵𝗮𝗿𝗻𝗲𝘀𝘀: loop, tools, context, permissions, memory. Vendors now ship the whole stack. Convenient, until you want to leave. You need to own the memory:   • Loops can be swapped. A year of corrections cannot.   • Readable text travels. Opaque features do not.   • OWASP ASI06: poisoned memory persists across sessions.   • Portable, inspectable, deletable, or you are renting it. What an agent harness is, and why memory is the part to own → neuraltrust.ai/blog/what-is-… #AgentHarness #AIMemory #AISecurity

Claude Duet works in the background with Gmail, Slack and Stripe access.   • 6,000+ connectors. Research preview, US Pro/Max only.   • Credentials stay off the model. Sends, pays and deletes need approval.   • No Team/Enterprise admin controls documented yet.   • Approval fatigue: only 13.6% of 1,053 testers refused a swapped dangerous command.   • Unmanaged agent until Anthropic ships the enterprise layer. What CISOs need to know → [neuraltrust.ai/blog/claude-d…](neuraltrust.ai/blog/claude-d…) #ClaudeDuet #AISecurity #EnterpriseAI

NeuralTrust has been positioned in the Catalysts quadrant of the ITMAP AI Gateways market map. Catalysts means strong Product Strategy and Execution. The vendors building the architecture the market will follow. Independent evaluation. No vendor funding. 12 vendors tracked. 👉 neuraltrust.ai/news/neuraltr… #AIGateway #AISecurity #CISO #NeuralTrust

NeuralTrust has been selected for @Siftedeu Rising AI 100 for 2026, the ranking of Europe’s leading AI companies!

Gartner also published the first-ever Hype Cycle for AI-Driven Cybersecurity today. NeuralTrust named a Sample Vendor in AI Runtime Defense, again. 58% of CISOs say GenAI delivers real innovation. 72% say it’s too expensive relative to the value. That gap is exactly what the Peak of Inflated Expectations looks like from the inside. #AIRuntimeDefense #AISecurity #Gartner #AgenticAI

Gartner published the first-ever Hype Cycle for Secure AI today. NeuralTrust named a Sample Vendor in AI Runtime Defense. Peak of Inflated Expectations. High benefit. Two numbers that matter: • 2028: skipping preproduction AI testing = 2x more incidents • 2027: task-driven agent abuses will cost 4x more than multi-agent systems The threat is more immediate than most teams assume. #AIRuntimeDefense #AISecurity #Gartner #AgenticAI

NeuralTrust named a Sample Vendor in the Gartner Emerging Tech Impact Radar: AI Cybersecurity, published today. AI Application Security. Range: 1–3 years. Mass: High. Gartner: "By 2029, at least 50% of enterprise AI security budgets will prioritize in-line runtime protection over general AI visibility, up from less than 10% today." $16.4B end-user spend forecast by 2030. #AIRuntimeDefense #AISecurity #Gartner #AgenticAI

Claude had 6 incidents in September. StatusGator has logged 202 since January. If your team stops every time Claude goes down, the problem is the architecture. 529 = Anthropic’s problem. 429 = usually yours. Different fixes. Status check, error guide, September incident timeline, and 5-step fallback setup → neuraltrust.ai/blog/is-claud… #Claude #AIGateway #LLMResilience

Google announced Gemini 4 Argon on September 30. - Vendor benchmarks: leads DeepSWE v1.1 and the Vals Index. - Artificial Analysis independent score: 52.6 vs 57.6 for Claude Opus 5.5. - Access: Fairwind cyber defenders only. No public API yet. Strong, uneven, and unavailable to most teams. Full breakdown → neuraltrust.ai/blog/gemini-4… #Gemini4 #AIModels #AISecurity #CISO

Most enterprises buying an AI gateway are evaluating the wrong things. Deployment model needs to be answered first. A cloud-only gateway cannot meet a UK GDPR or EU AI Act self-hosted requirement, regardless of the feature list. 7 criteria. A vendor scorecard. 10-step evaluation process. The red flags that disqualify before the demo. Enterprise buyer's guide → neuraltrust.ai/blog/how-to-c… #AIGateway #EnterpriseAI #CISO

Anthropic released Claude Opus 5.5 today. 85% fewer sandbox escape attempts. Lowest prompt injection rate tested. 40% lower cost than Opus 5. But model-level security and enterprise runtime protection are two different things. What Opus 5.5 doesn’t include, and how enterprises close the gap → neuraltrust.ai/blog/claude-o… #ClaudeOpus55 #AISecurity #AgenticAI #CISO

The NCSC published new agentic AI guidance in August 2026. It opens by acknowledging real incidents had already occurred: agents accessing data they should not have, taking actions outside their defined scope. 7 considerations. 4-level sandboxing maturity model. 24/7 SOC monitoring required. Full breakdown → neuraltrust.ai/blog/ncsc-ai-… #NCSC #AgenticAI #AISecurity #CISO

AI incidents: near zero in 2012. 223 in 2024. And that curve was built on chatbots and model APIs. AI agents are only just hitting production at scale. We mapped what that means for security teams: download the guide for free → neuraltrust.ai/guides/agenti… #AgenticAI #AISecurity #CISO

NeuralTrust is at the Gartner Security & Risk Management Summit in London this week: 22–24 September. Come find us if you want to talk AI agent security in production. #GartnerSRM #AISecurity #CISO #NeuralTrust

UK CISOs face AI security pressure from three directions at once. NCSC guidance. UK GDPR. EU AI Act, which still applies if you have EU customers. And the threat surface underneath all three is one that firewalls and endpoint tools were never designed to see. We published the complete enterprise AI security guide for UK CISOs → neuraltrust.ai/blog/ai-secur… #AISecurity #CISO #NCSC #UKGDPR

Most enterprises have an AI policy. Very few have AI governance that actually works in production. The difference: a policy is a document. Governance is the control layer that enforces it on every model call, in real time. Gartner: by 2028, loss of control over AI agents will be the top concern for 40% of Fortune 1000 orgs. We published the full guide → neuraltrust.ai/blog/ai-gatew… #AIGovernance #EUAIAct #AIGateway #CISO

We are happy to announce we have been named a Pioneer in the Gartner Emerging Market Quadrant for AI Application Security. First-ever Gartner quadrant for this category. Best-positioned European platform in the Pioneer quadrant. Full announcement: neuraltrust.ai/news/neuraltr… Gartner announcement: gartner.com/document-reader/… #AIApplicationSecurity #Gartner