[ExI] OpenAI Says It Will Not Release Newest A.I. Model Over Safety Concerns

[Jason Resch]

On Tue, Sep 29, 2026 at 10:18 AM John Clark <johnkclark at gmail.com> wrote:

>
>
> On Tue, Sep 29, 2026 at 9:35 AM Jason Resch via extropy-chat <
> extropy-chat at lists.extropy.org> wrote:
>
> *> They've been using this same excuse since GPT 2.*
>>
>
> *And the smarter their AIs have become the louder their warnings have
> become.  *
>


For context, I wrote this
<https://alwaysasking.com/when-will-ai-take-over/#Comprehension> in 2020:

In 2019, engineers at OpenAI <https://en.wikipedia.org/wiki/OpenAI> succeeded
to such a degree that they were scared by their creation. They considered
their AI so dangerous they said it would be irresponsible to release to the
public.

These engineers created an AI called *Generative Pretrained Transformer 2*
 or GPT-2 <https://en.wikipedia.org/wiki/OpenAI#GPT-2>.

[snip]

The creators feared their technology would be exploited to create fake news
or generate spam.

Tristan Greene <https://www.bbc.com/news/technology-49446729> said, “I’m
terrified of GPT-2 because it represents the kind of technology that evil
humans are going to use to manipulate the population – and in my opinion
that makes it more dangerous than any gun.”

After nine months of debate between OpenAI and the broader community,
OpenAI reversed its decision. GPT-2 is now available to the public for
anyone to download and use.




> * > Yet it always comes out. I think it's just their marketing strategy.*
>>
>
> *Could you clarify what you mean by that? *
>

It excites their investors to hear "Our AI is so good, we can't even
release it." And it is also a convenient catch-all excuse when they run
behind their planned release schedule.


> *Regardless of what the motivations of AI companies are for saying what
> they're saying, do you think AI is unlikely to produce changes to our
> civilization of unprecedented size?  *
>

Where did it appear I said that? Have you not seen my Singularity talk
slides
<https://docs.google.com/presentation/d/1bITq1_XCNY4sEmZhbkcRUr77PtWawP3rpLxNYZxDTBY/edit?usp=sharing>
or presentation <https://youtu.be/KUPTtraCK-g>? We've already witnessed
unprecedent changes. But the crying wolf warnings for the past 6 years
where they say, "It's too dangerous, we can't release it," and then
inevitably release it they're are doing a disservice to those trying to get
things done for AI safety. The neuralese thing you posted in the other
thread is a perfect example. Those who work in AI safety generally want
this avenue banned as it prevents them from testing models for safety and
alignment.

Jason


>
>
>>
>> On Tue, Sep 29, 2026, 7:55 AM John Clark via extropy-chat <
>> extropy-chat at lists.extropy.org> wrote:
>>
>>> Explore this gift article from The New York Times. You can read it for
>>> free without a subscription.
>>>
>>> OpenAI Says It Will Not Release Newest A.I. Model Over Safety Concerns
>>>
>>> The company’s researchers raised questions about the security of the
>>> model, known as GPT-6.1 Astra.
>>>
>>>
>>> https://www.nytimes.com/2026/09/28/technology/openai-astra-safety.html?unlocked_article_code=1.E1E.s0GM.WjUmxr83iqqM&smid=em-share
>>>
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20260929/a8f17c55/attachment.htm>