- Mindgard has raised $30 million in Sequence A funding, bringing its whole funding to virtually $42 million.
- The platform has discovered greater than 150 vulnerabilities in AI merchandise, together with a zero-day in Cursor and weaknesses in ChatGPT’s guardrails.
- Two main rivals had been acquired up to now yr, and the AI safety market is anticipated to succeed in $56.5 billion by 2033.
Researchers discovered a method to make Cursor’s AI-powered editor run any code while not having phishing hyperlinks or malware. The identical group additionally leaked Google Antigravity knowledge throughout belief boundaries and bypassed ChatGPT’s image-generation guardrails.
Mindgard disclosed all three vulnerabilities.
The startup accomplished a $30 million Sequence A spherical led by Album VC, with new investor Karma Ventures and returning traders .406 Ventures, Atlantic Bridge, IQ Capital, and Lakestar. This brings its whole funding to just about $42 million, up from about $11.9 million raised in a 2023 seed spherical and a December 2024 Series A tranche.
“AI is creating a completely new assault floor, and organisations want a essentially completely different strategy to securing it. We don’t simply automate assaults. We operationalize experience, turning the data of main AI safety researchers and offensive safety practitioners into the capabilities each enterprise must safe their AI,” mentioned James Brear, chief government of Mindgard.
From a Lancaster College lab to a industrial red-teaming platform
Based in 2022, the startup started in what it describes because the world’s largest AI safety lab, constructed on greater than ten years of analysis by Dr. Peter Garraghan at Lancaster College. Garraghan is now Mindgard’s chief science officer.
James Brear, who beforehand led Swimlane, Veriflow, and Procera, grew to become CEO in October 2025 as the corporate moved from analysis to enterprise gross sales. The corporate has not shared its present headcount.
The platform runs ongoing, automated red-teaming in opposition to fashions, brokers, and functions, simulating actual attacker conduct quite than counting on static checklists. It does three important issues: finds unmonitored “shadow AI,” assessments deployed methods for vulnerabilities, and screens AI functions in operation. For instance, a financial institution launching a buyer chatbot might use the platform to verify for immediate injection and knowledge leaks earlier than and after launch.
Mindgard stands out by providing proof of labor, with greater than 150 disclosed vulnerabilities in actual merchandise added to its menace database. The corporate works with shoppers in monetary companies, prescription drugs, gaming, digital companies, semiconductors, and healthcare.
Quickly consolidating and well-funded class
The AI safety sector has shifted from a crowded market to a consolidating one inside a yr. HiddenLayer raised $56 million, backed by Microsoft’s M12 and IBM Ventures. Zurich-based Lakera, recognized for its Gandalf prompt-injection sport, raised $30 million earlier than being acquired by Test Level in September 2025. Defend AI secured $60 million at a reported $400 million valuation earlier than its acquisition by Palo Alto Networks the identical yr.
Whereas different firms deal with guardrails, monitoring, or posture administration, Mindgard prioritises offensive analysis by specializing in discovering new exploits. This technique is tougher to repeat, but it surely means Mindgard should keep forward of attackers who even have entry to AI instruments. Two of its closest rivals, which focus on endpoint safety and stopping AI-era threats, have also raised large amounts of funding this year, displaying how aggressive this sector is.
“Organisations are transferring AI into essential operations with out safety infrastructure designed for a way these methods function in observe. Mindgard has translated deep analysis and elite offensive-security experience right into a repeatedly evolving functionality that allows enterprises to remain forward of rising AI threats,” mentioned Ty Boswell, associate at Album VC.
“AI safety is an rising class, and Mindgard stands out as one of many names to look at. Its expertise is constructed on world-class offensive safety analysis, combining automated reconnaissance with sensible penetration testing. Crucially, it’s already confirmed in manufacturing — a number of the most demanding F2000 enterprises depend on it to safeguard their AI methods,” added Kristjan Laanemaa, founding associate at Karma Ventures.
The long run past vulnerability discovery
The AI safety platforms market was valued at about $15.8 billion this year and is anticipated to succeed in $56.5 billion by 2033. The smaller however faster-growing agentic AI safety section might develop greater than ten occasions by the early 2030s, in keeping with some estimates.
The brand new funding will assist develop product improvement, engineering, gross sales, and advertising and marketing.
Lakera and Defend AI have been purchased and at the moment are a part of larger safety platforms. Mindgard, HiddenLayer, and some others are nonetheless competing for market share. Now, the principle query is who can flip these discoveries into long-term enterprise contracts earlier than an even bigger firm buys them.
