OpenAI alignment researcher leaves to build non-invasive mind-reading AI

Naomi Bashkansky, a researcher who previously worked in OpenAI's alignment division, has announced that she is leaving the company to build AI capable of reading human thoughts non invasively. The announcement first surfaced in a post on Polymarket's X account, which cited a statement from Bashkansky, and was later reinforced by a personal essay on her website titled "Why I'm leaving OpenAI to build telepathy." Her departure comes amid growing attention to AI safety issues and the research direction taken by OpenAI. Bashkansky, who focused on aligning AI behavior with human values, is now shifting to developing AI based telepathy technology, a field with the potential to transform how humans interact with machines and each other.

In her essay, Bashkansky explained her motivation for building technology that can read thoughts non invasively. She emphasized the importance of this approach compared to invasive methods that require implanting devices in the brain. Non invasive technology, she argued, could pave the way for more natural and widely accessible communication, without the risks of surgery or long term side effects. Bashkansky's move does not happen in a vacuum. Previously, several key OpenAI researchers, including Ilya Sutskever and Jan Leike, had left the company, raising questions about the future of AI safety research within the organization. Jan Leike, who once led the alignment team, previously explained OpenAI's efforts to protect humanity from superintelligent AI, as reported by IEEE Spectrum.

The departure of these figures indicates a shift in priorities within OpenAI, which is increasingly focused on developing commercial products. Bashkansky herself is not a big name in the public sphere like Sutskever or Leike, but her departure is still significant because she represents a wave of researchers choosing to pursue their visions outside the structure of large companies. Her focus on non invasive telepathy shows that part of the AI research community sees great potential in brain computer interfaces as the next evolution of human machine interaction. From an industry perspective, this move can be seen as a signal that AI research is not only centered on large language models or general intelligence, but also on more personal and profound applications such as mind reading.

Such technology, if successfully developed, could have broad implications for healthcare, communication, and even cybersecurity. However, it is important to note that claims about non invasive mind reading capabilities are still conceptual, and no concrete product has been announced. Polymarket, as a prediction market platform, only served as an intermediary for information in this announcement. The main event to watch is Bashkansky's decision to leave OpenAI and establish a new venture. No further details are available regarding funding, partners, or the development timeline for the technology in the available sources. The broader context shows that the movement of researchers from OpenAI to independent projects is not new. Sutskever, for example, after leaving OpenAI, founded his own company focused on AI safety.

This pattern indicates that although OpenAI is a leading center for AI development, some of its best talent chooses to pursue research that is more mission oriented, including high risk areas like telepathy. In her essay, Bashkansky did not provide technical details on how she would build a non invasive mind reading AI. However, she emphasized that the approach would leverage brain signals detectable from outside the skull, such as electroencephalography (EEG) or functional near infrared spectroscopy (fNIRS), which have been used in brain computer interface research for years. Nevertheless, these claims cannot be verified from available sources. The long term impact of Bashkansky's departure on OpenAI remains uncertain.

But symbolically, it adds to a long list of researchers leaving the company due to differences in vision or a desire to explore more niche areas. For the AI industry, this phenomenon shows that the research ecosystem is becoming increasingly fragmented, with experts spreading in various directions, from AI safety to more experimental applications like telepathy. There has been no official response from OpenAI regarding Bashkansky's departure. Meanwhile, Bashkansky appears set to begin her new journey with a focus on developing technology that could revolutionize the way we communicate. If successful, this move could place her at the forefront of a new wave of human centric AI innovation.

Industry observers will be watching the next developments in Bashkansky's telepathy project, especially whether she can attract funding and build a team capable of realizing her vision. In the meantime, her departure serves as a reminder that the world of AI research is never static, and researchers continue to push the boundaries of what is possible, both inside and outside large companies.