What happened
Bloomberg reports that OpenAI president and co-founder Greg Brockman said he believes artificial general intelligence has already been achieved. He referred to an earlier definition of AGI as autonomous computer systems that outperform humans on most economically valuable tasks. The source does not provide independent evidence that this threshold has been met.
Bloomberg reports that Greg Brockman made the remarks during a media briefing OpenAI held the previous week to present its GPT-6 Astra model. According to Bloomberg, Brockman said that AGI, or artificial general intelligence, had already been achieved.
The report says Brockman invoked a definition he had previously used: autonomous computer systems that exceed human performance on most economically valuable tasks. Bloomberg presents the current status of AGI as ambiguous and says the claim can sound like vague marketing.
The source contains no independent testing, benchmark data, technical documentation or outside expert assessment confirming that GPT-6 Astra or another system satisfies that definition. It also does not describe a change to access, pricing or availability for GPT-6 Astra.
Source details: bloomberg.com ↗
Why it matters
The statement matters because it comes from a senior OpenAI executive and could influence how companies, policymakers and the public interpret the significance of GPT-6 Astra and current AI capabilities. But the claim is not an independently verified finding. Bloomberg’s report provides no benchmark results, external evaluation, deployment evidence or agreed definition establishing that AGI has arrived. The practical meaning therefore remains uncertain.
A senior executive’s public characterization of AGI is consequential because it can shape expectations about the capabilities and economic effects of frontier models. It also adds another executive-level claim to the continuing public debate surrounding GPT-6 Astra.
The evidentiary limitation is substantial. Bloomberg reports Brockman’s assertion, but the source does not independently confirm it. A declaration that AGI has arrived is not itself evidence that a system can reliably outperform humans across most economically valuable tasks.
The comment may also highlight a definitional problem: the threshold described by Bloomberg is broad, and the source does not specify which tasks, comparison groups, reliability standards or time horizons would be used to assess it.
What to watch next
Watch for technical evidence, independent evaluations and clearer definitions from OpenAI or outside researchers. The source does not report a new product-access change, pricing change or general availability decision tied to Brockman’s comment. It also does not establish whether OpenAI’s claim reflects an official company position or Brockman’s personal view.
OpenAI’s publication of reproducible evaluations or detailed evidence would be more informative than an executive characterization alone. Independent testing would be needed to assess whether claimed capabilities generalize beyond selected demonstrations or internal assessments.
Readers should distinguish the reported statement from a confirmed product milestone. Bloomberg does not report that GPT-6 Astra is generally available, what it costs, or that users have received access because of this announcement.
Further reporting may clarify whether Brockman’s wording represents an official OpenAI definition or policy, and whether other researchers accept the performance threshold he referenced.