In a market where the intro\u00adduc\u00adtion of AI is progres\u00adsing rapidly, the Israeli startup Irre\u00adgu\u00adlar has now closed a finan\u00adcing round of 80 million US dollars. The inves\u00adtors include Sequoia Capi\u00adtal, Redpoint Ventures, Swish Ventures, Assaf Rappa\u00adport and Ofir Ehrlich. The company, foun\u00added by Dan Lahav and Omer Nevo, specia\u00adli\u00adzes in stress test\u00ading advan\u00adced AI models under real-world condi\u00adti\u00adons, working with leading AI labs such as OpenAI and Anthropic. <\/p>\n
Irre\u00adgu\u00adlar was foun\u00added in 2023 by Dan Lahav<\/strong> and Omer Nevo<\/strong>. Lahav, who previously worked at LabPi\u00adxies, a startup acqui\u00adred by Google, and later worked as an AI rese\u00adar\u00adcher at IBM, brings exten\u00adsive expe\u00adri\u00adence in AI rese\u00adarch. Nevo is a serial entre\u00adpre\u00adneur with a back\u00adground in cyber\u00adse\u00adcu\u00adrity. Toge\u00adther, they have built a company that focu\u00adses on evalua\u00adting and secu\u00adring advan\u00adced AI models. The new funding will be used to expand rese\u00adarch and deve\u00adlo\u00adp\u00adment, expand the team, inten\u00adsify colla\u00adbo\u00adra\u00adtion with leading AI labs <\/p>\n Busi\u00adness model<\/strong><\/p>\n Irre\u00adgu\u00adlar\u2019s busi\u00adness model is based on conduc\u00adting stress tests for AI systems under real threat scena\u00adrios. This includes tests that focus on how AI models react to attempts to circum\u00advent anti\u00advi\u00adrus soft\u00adware or carry out auto\u00adno\u00admous offen\u00adsive operations. <\/p>\n The aim is to iden\u00adtify poten\u00adtial vulnerabi\u00adli\u00adties and deve\u00adlop solu\u00adti\u00adons to ensure the secure imple\u00admen\u00adta\u00adtion of these systems on a large scale.<\/p>\n The foun\u00adders explain: \u201cIt\u2019s about prac\u00adti\u00adcal tools that stop thre\u00adats where\u00adver AI is deve\u00adlo\u00adped or used.\u201d<\/p>\n The start-up\u2019s tech\u00adno\u00adlogy has alre\u00adady attrac\u00adted the atten\u00adtion of leading AI labs: OpenAI uses Irre\u00adgu\u00adlar\u2019s ratings in system maps for GPT\u20113, GPT\u20114 mini and GPT\u20115. Irre\u00adgu\u00adlar has also co-autho\u00adred a white paper with Anthro\u00adpic on confi\u00adden\u00adtial infe\u00adrence systems, while rese\u00adar\u00adchers at Google Deep\u00adMind have cited the company\u2019s plat\u00adform in studies on AI cyber\u00adat\u00adtack capabilities.<\/p>\n The funding reflects a broa\u00adder realiza\u00adtion across the tech\u00adno\u00adlogy land\u00adscape that the success of AI depends not only on buil\u00adding ever more powerful models, but also on protec\u00adting them from misuse, acci\u00addents and new threat vectors. As compa\u00adnies and govern\u00adments acce\u00adle\u00adrate adop\u00adtion, the demand for proac\u00adtive AI secu\u00adrity has increased. With its strong inves\u00adtor support and early trac\u00adtion, Irre\u00adgu\u00adlar is posi\u00adtio\u00adning itself as a funda\u00admen\u00adtal player in this emer\u00adging layer of AI infrastructure. <\/p>\n <\/p>\n <\/p>\n","protected":false},"excerpt":{"rendered":" In a market where the intro\u00adduc\u00adtion of AI is progres\u00adsing rapidly, the Israeli startup Irre\u00adgu\u00adlar has now closed a finan\u00adcing round of 80 million US dollars. The inves\u00adtors include Sequoia Capi\u00adtal, Redpoint Ventures, Swish Ventures, Assaf Rappa\u00adport and Ofir Ehrlich. The company, foun\u00added by Dan Lahav and Omer Nevo, specia\u00adli\u00adzes in stress test\u00ading advan\u00adced AI [\u2026]<\/p>\n","protected":false},"author":2,"featured_media":56913,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"wp_typography_post_enhancements_disabled":false,"footnotes":""},"categories":[1366],"tags":[],"class_list":["post-56915","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news-en"],"yoast_head":"\n