Patronus AI
Patronus AI is a technology company founded in 2023 that develops tools for evaluating and monitoring artificial intelligence (AI) systems. The company's tools aim to address issues such as hallucinations, copyright violations, and safety risks in AI systems.[1][2]
Patronus AI was founded by Anand Kannappan and Rebecca Qian to improve the safety and transparency of AI systems.[3]
The company has received investments, including a $3 million seed funding round,[4] and also a $17 million funding round to support its work in mitigating AI-related risks.[5][6]
Industry impact[edit]
Patronus AI's developments for evaluating AI technologies, particularly in regulated sectors have been recognized and awarded.[7] On July 11, 2024, Patronus AI was awarded the "Best Enterprise Implementation of Generative AI (Finance)", during 6th annual AI innovation awards presented by VentureBeat.[8] Patronus AI has also been used to test models like ChatGPT and Claude for displays of potential copyright infringement.[9]
References[edit]
- ^ Nuñez, Michael (2024-07-11). "Meet Patronus AI's 'Lynx': The open-source bullshit detector outsmarting GPT-4". VentureBeat. Retrieved 2024-07-12.
- ^ "A Patronus AI apresenta uma ferramenta para identificar erros de IA generativa". contxto.com. Retrieved 2024-07-12.
- ^ Miller, Ron (2023-09-14). "Patronus AI conjures up an LLM evaluation tool for regulated industries". TechCrunch. Retrieved 2024-07-12.
- ^ "Patronus AI rolls out SimpleSafetyTests to identify harmful AI models". SiliconANGLE. 2023-12-19. Retrieved 2024-07-12.
- ^ Nuñez, Michael (2024-05-22). "Patronus AI secures $17M to tackle AI hallucinations and copyright violations, fuel enterprise adoption". VentureBeat. Retrieved 2024-07-12.
- ^ Shrivastava, Rashi. "This AI-Powered 'Coach' Catches Hallucinations In Other AI Models". Forbes. Retrieved 2024-07-12.
- ^ Woodie, Alex (2023-10-23). "AI Needs a Third-Party Benchmark. Will It Be Patronus AI?". Datanami. Retrieved 2024-07-12.
- ^ Mohammed, Aliyah (2024-07-11). "Announcing the winners of VentureBeat's 6th Annual AI Innovation Awards". VentureBeat. Retrieved 2024-07-12.
- ^ Field, Hayden (2024-03-06). "Researchers tested leading AI models for copyright infringement using popular books, and GPT-4 performed worst". CNBC. Retrieved 2024-07-12.