OPENAI INSIDERS LEAK SECRET AGI TIMELINE: "IT WILL HAPPEN IN 18 MONTHS"
Three former OpenAI engineers have come forward with explosive claims that the company's internal projections show artificial general intelligence arriving far sooner than any public statement has suggested. Speaking under condition of anonymity, the whistleblowers allege that an internal document — codenamed Project Threshold — outlines a pathway to AGI within 18 months, contingent on next-generation hardware shipments from a major chip manufacturer.
The leaked document, portions of which were shared with Exposed Daily through an encrypted channel, describes a system codenamed Q* (pronounced "Q-Star") that allegedly passed a battery of cognitive benchmarks designed to simulate human-level reasoning across domains including mathematics, legal analysis, and scientific hypothesis generation.
"We were told not to discuss these results externally," said one source, who claims to have been present at an all-hands meeting where the findings were presented. "The gap between what they say publicly and what they know internally is enormous."
OpenAI declined to comment specifically on the existence of Project Threshold, issuing only a boilerplate statement affirming their commitment to "responsible AI development." However, internal Slack screenshots shared with this outlet appear to show senior research staff discussing "containment protocols" in the event of what one message calls "capability emergence."
Critics within the AI safety community say the allegations — if true — represent a catastrophic breach of the transparency norms the company publicly champions. "If they're hiding timelines like this from regulators and the public while continuing to train more powerful systems, we have a systemic accountability problem," said one prominent AI safety researcher.
The sources allege that board members have been briefed on the accelerated timeline but that investor pressure to maintain competitive parity with Google DeepMind and Meta AI has created internal pressure to keep the information classified. One source claims that at least two senior safety researchers resigned in the past quarter specifically over this issue.
Exposed Daily has submitted a formal inquiry to OpenAI's communications team and will update this story as new information becomes available. The ref document hash, provided by one source for verification purposes, has been submitted to a blockchain timestamp service.
IS THIS REAL?
STAY IN THE LOOP
Get breaking leaks before the mainstream catches on. No spam. Unsubscribe anytime.
THIS STORY IS GOING VIRAL ON TIKTOK
Share this leak before it gets taken down.