Prompt Security
Definition
Prompt Security is an Israeli company founded in 2023 that developed security designed specifically for business use of language models. It was acquired by SentinelOne in 2025 for about $250 million.
Two security gaps that opened up with GenAI
The company was founded in August 2023 by Itamar Golan (CEO) and Lior Drihem (CTO), both veterans of Unit 8200 who had previously worked on AI at cyber companies including Check Point and Orca Security. It was founded around two new risks. The first is "shadow AI": sensitive company information pasted into an outside chat, without the IT department even knowing the tool is in use. The second is prompt injection, in which text hidden in a document makes the model ignore the instructions it was given and follow an attacker's instructions instead.
The platform: a middle layer that checks every prompt and every answer
The company's platform checked in real time every prompt sent to the model and every answer that came back from it, in order to block data leaks, harmful content and attacks designed specifically for the world of models. The company raised $5 million in a seed round in January 2024 and $18 million in a Series A round in November of the same year, with participation from technology companies such as Okta and F5.
The acquisition by SentinelOne, according to the buyer's official announcement
On 5 August 2025 SentinelOne officially announced that it had signed a definitive agreement to acquire Prompt Security. According to the announcement, the deal was paid in a combination of cash and stock and was expected to close in the third quarter of SentinelOne's fiscal year 2026. The deal was in fact completed in September 2025. Itamar Golan, CEO and co-founder of Prompt Security, was quoted in the announcement: "As enterprise adoption of GenAI and agentic AI accelerates, the security and privacy risks are rapidly shifting from theoretical to operational."
What SentinelOne got in the deal
According to the official announcement, Prompt Security's technology includes MCP (Model Context Protocol) gateway security between AI applications and more than 13,000 known MCP servers, and model-agnostic coverage of all the major AI providers (OpenAI, Anthropic and Google) alongside self-hosted models. Tomer Weingarten, SentinelOne's CEO, described it as "the foundation for secure AI adoption at scale."
Why a two-year-old company sells for $250 million
The deal, valued at about $250 million according to a Calcalist report, opened a wave of acquisitions in AI security in Israel. The reason for the high price relative to the company's age is not only the technology but the timing: established cyber corporations were looking for a fast way into a field that had opened up in front of them without warning.
FAQ ❓
Who acquired Prompt Security, and when?
SentinelOne. The definitive agreement was signed and announced on 5 August 2025, according to SentinelOne's official announcement, and the deal was in fact completed in September 2025.
Who founded Prompt Security?
Itamar Golan (CEO) and Lior Drihem (CTO), in August 2023. Both are veterans of Unit 8200.
What is prompt injection?
Text hidden in a document, a web page or an email that makes a language model ignore the instructions it was originally given and follow the attacker's instructions instead. It is a structural weakness of models that cannot tell the difference between an instruction and the content they are reading.
How much did SentinelOne pay for Prompt Security?
According to a Calcalist report, about $250 million, in a combination of cash and stock according to SentinelOne's official announcement. The exact amount was not officially disclosed by the buyer.
Found a mistake, or have a source we should add? Write to us