Alex Bilzerian

Alex Bilzerian
American
Bilzerian winning 1st Prize in the OpenOcean Poker Tournament of 2021 in Helsinki
Born	(1996-08-05) 5 August 1996 Worcester, MA, U.S.
House	Artsruni dynasty
Education	Massachusetts Academy of Math and Science at WPI
Alma mater	Cornell University (BA) Worcester Polytechnic Institute

Alexander John Bilzerian (born 5 August 1996) is an American physicist, AI engineer and poet with a background in venture capital. After serving as Machine Learning Lead at Hive AI (mostly supervised learning and computer vision) through the company's growth stage, he was tapped by MySQL’s founder and author Michael Widenius to develop OpenOcean’s AI investment thesis for his venture firm in Finland. Bilzerian made pioneering contributions to Large Language Model red-teaming techniques in the prompt injection space while solo-developing an open-source toolkit, the "Prompt Library" for prompt engineering. His work concerns artificial intelligence, cognition, human–computer interaction, information, probability and poetry.

Early life and Education

Alex was raised in Massachusetts. He attended the Imus Ranch several times in his youth, personally riding between the mesas of New Mexico with Don Imus on horseback. His cousin Paul Bilzerian is a former corporate takeover specialist. Another cousin is a poker player, Dan Bilzerian. Alex was educated at the Massachusetts Academy of Math and Science at WPI, one of the top specialized programs in the United States, where he conducted research at Worcester Polytechnic Institute on Tumor Treating Fields—a form of directed energy—and RNA interference for genetics research related to cancer treatment in 2014. That same year, he launched an engineering initiative to help a local boy with a rare partial blindness condition play basketball again with an optical head-mounted display to identify blind spots, a computer vision approach — Bilzerian's first human-computer interaction project.

He studied Physics and Creative Writing at Cornell University after spending a full year at Worcester Polytechnic Institute in lieu of traditional high school. During his studies, he received the George Harmon Coxe Poetry Award from Cornell University's Department of English.

Career

Bilzerian served as Machine Learning Lead at Hive AI under their content moderation division from 2019 to 2021, where his work concerned computer vision systems for commercial clients like Reddit, Amazon (company), and TimeWarner.

In addition to authoring OpenOcean's AI investment thesis, he also made deals, conducted due diligence, and offered his guidance to the firm when it came to their partnership with the University of Oxford and the Saïd Business School for their investment committee's inference, infrastructure, and AI needs during the collaboration. After the success of the Helsinki job, Bilzerian briefly became a VC Investor at a Silicon Valley VC firm in 2022 before he spun off his own operation to focus on his research programme encompassing AI safety, AI ethics, AI privacy, red teaming, and prompt injection. Apart from his open-source success, his early conceptual work on LLM-based security vulnerabilities, disclosed in 2022, contributed to and influenced the broader research landscape with respect to AI's benefits and risks. He is cited in a variety of academic publications on cybersecurity vulnerabilities and in a book on knowledge management. When Zimmerman and Zollikofer approached Bilzerian for direction on their research into AI, he provided generous advice and spoke to them personally. They won the Swiss AI Safety Award (1st Place) for their paper, "Synthetic Cancer" (2024), which demonstrated metamorphic malware able to leverage LLMs for code generation and social engineering. The authors, in their work, gave Bilzerian a nod for being among the first to prototype LLM-based worms, identifying and looking to counter the emerging threat early.

Otter.ai Privacy Disclosure

In September 2024, Bilzerian disclosed on X (formerly Twitter) that Otter.ai had inadvertently sent him a transcript containing lengthy confidential conversations between venture capitalists that occurred after he had left a Zoom call. His post went viral, amassing over 6 million views and garnering significant global and domestic media coverage from publications like the Washington Post and Entrepreneur Magazine.

The incident was documented in the AI Incident Database as Incident #811 and was covered by many international media outlets, including Fortune, The New York Post, India Today, and the New Zealand Herald.

The disclosure was analyzed by legal scholars at the University of Wisconsin-Madison concerning best practices and compliance risks for organizations using AI note-taking tools. Legal and compliance publications highlighted the incident as a significant case study in AI privacy vulnerabilities.

In August 2025, a class-action lawsuit (Brewer v. Otter.ai, Case No. 5:25-cv-06911) was filed in the United States District Court for the Northern District of California. They appreciated Bilzerian's disclosure—directly citing it—given the significance of the revelations. Otter.ai is alleged to have violated the Electronic Communications Privacy Act, the Computer Fraud and Abuse Act, and the California Invasion of Privacy Act. The lawsuit alleges that they failed to obtain informed consent from all meeting participants before recording conversations and using the transcripts for AI model training purposes.