Anthropic, OpenAI's proposed AI risk evaluators may not have enough power to prevent disasters
Published . Updated .
Anthropic and OpenAI are proposing embedded AI evaluators to help manage risk of models causing catastrophic harm to society, but the idea has some issues.
Directly affected
- No company is named in this report.
Indirectly affected
- SUPX SUPERX AI TECHNOLOGY LTD
GA selected SUPX as a RELATED security because its canonical security name directly matches this event's themes: technology, ai. This is an explainable DERIVED market relationship, not a claim that the security was named in the source article.
- ABAT AMERICAN BATTERY TECHNOLOGY
GA selected ABAT as a RELATED security because its canonical security name directly matches this event's themes: technology. This is an explainable DERIVED market relationship, not a claim that the security was named in the source article.
- ACM AECOM TECHNOLOGY CORP
GA selected ACM as a RELATED security because its canonical security name directly matches this event's themes: technology. This is an explainable DERIVED market relationship, not a claim that the security was named in the source article.
- AOSL ALPHA & OMEGA SEMICONDUCTOR LT SHS
GA selected AOSL as a RELATED security because its canonical security name directly matches this event's themes: semiconductor. This is an explainable DERIVED market relationship, not a claim that the security was named in the source article.
- BBAI BIGBEAR AI HLDGS INC
GA selected BBAI as a RELATED security because its canonical security name directly matches this event's themes: ai. This is an explainable DERIVED market relationship, not a claim that the security was named in the source article.
- CHAT Roundhill Generative AI & Tech
GA selected CHAT as a RELATED security because its canonical security name directly matches this event's themes: ai. This is an explainable DERIVED market relationship, not a claim that the security was named in the source article.
Price reaction, T-5 to T+3
Anthropic and OpenAI are proposing embedded AI evaluators to help manage risk of models causing catastrophic harm to society, but the idea has some issues.
Source: CNBC Top
How far SuperX AI Technology (SUPX) has moved since 16 Sep 2026
Since the first closing price after “Anthropic, OpenAI's proposed AI risk evaluators may not have enough power to prevent disasters,” SuperX AI Technology (SUPX) is -2.82%. The S&P 500 fund (SPY) is +0.52% over the same days, a -3.34% difference versus the market.
Filings that name SuperX AI Technology (SUPX) around 16 Sep 2026
- Donald J Trump buy Aecom (ACM) ($1,001 - $15,000). That is 19 days before publication, inside 30 days. The filing is listed beside the story “Anthropic, OpenAI's proposed AI risk evaluators may not have enough power to prevent disasters.”
- Donald J Trump sell Aecom (ACM) ($1,001 - $15,000). That is 23 days before publication, inside 30 days. The filing is listed beside the story “Anthropic, OpenAI's proposed AI risk evaluators may not have enough power to prevent disasters.”
Public disclosure filings, not a recommendation. The filing link is the source document. News use of these reports still sits under 5 U.S.C. § 13107(c).
Who gains and who loses around SuperX AI Technology (SUPX)
Companies carried with this story
- SUPX SuperX AI Technology
In “Anthropic, OpenAI's proposed AI risk evaluators may not have enough power to prevent disasters,” SuperX AI Technology (SUPX) is linked, with no stored direction, medium confidence. Reason on file: GA selected SUPX as a RELATED security because its canonical security name directly matches this event's themes: technology, ai. This is an explainable DERIVED market relationship, not a claim that the security was named in the source article
- ABAT American Battery Technology
In “Anthropic, OpenAI's proposed AI risk evaluators may not have enough power to prevent disasters,” American Battery Technology (ABAT) is linked, with no stored direction, medium confidence. Reason on file: GA selected ABAT as a RELATED security because its canonical security name directly matches this event's themes: technology. This is an explainable DERIVED market relationship, not a claim that the security was named in the source article
- ACM Aecom
In “Anthropic, OpenAI's proposed AI risk evaluators may not have enough power to prevent disasters,” Aecom (ACM) is linked, with no stored direction, medium confidence. Reason on file: GA selected ACM as a RELATED security because its canonical security name directly matches this event's themes: technology. This is an explainable DERIVED market relationship, not a claim that the security was named in the source article
- AOSL Alpha & Omega Semiconductor
In “Anthropic, OpenAI's proposed AI risk evaluators may not have enough power to prevent disasters,” Alpha & Omega Semiconductor (AOSL) is linked, with no stored direction, medium confidence. Reason on file: GA selected AOSL as a RELATED security because its canonical security name directly matches this event's themes: semiconductor. This is an explainable DERIVED market relationship, not a claim that the security was named in the source article
- BBAI BigBear.ai Holdings
BigBear.ai Holdings (BBAI) is linked, with no stored direction, medium confidence in the 16 Sep 2026 story “Anthropic, OpenAI's proposed AI risk evaluators may not have enough power to prevent disasters.” Reason on file: GA selected BBAI as a RELATED security because its canonical security name directly matches this event's themes: ai. This is an explainable DERIVED market relationship, not a claim that the security was named in the source article
- CHAT Roundhill Generative AI & Tech
Roundhill Generative AI & Tech (CHAT) is linked, with no stored direction, medium confidence in the 16 Sep 2026 story “Anthropic, OpenAI's proposed AI risk evaluators may not have enough power to prevent disasters.” Reason on file: GA selected CHAT as a RELATED security because its canonical security name directly matches this event's themes: ai. This is an explainable DERIVED market relationship, not a claim that the security was named in the source article
SuperX AI Technology (SUPX)
On 16 Sep 2026, “Anthropic, OpenAI's proposed AI risk evaluators may not have enough power to prevent disasters” is the story that connects SuperX AI Technology (SUPX) to this story.
What ties SuperX AI Technology (SUPX) to the headline is this: GA selected SUPX as a RELATED security because its canonical security name directly matches this event's themes: technology, ai. This is an explainable DERIVED market relationship, not a claim that the security was named in the source article
A second company, American Battery Technology (ABAT), is listed for a different reason: GA selected ABAT as a RELATED security because its canonical security name directly matches this event's themes: technology. This is an explainable DERIVED market relationship, not a claim that the security was named in the source article
Dates in the story, and the SuperX AI Technology (SUPX) price range
ABAT traded between 1.42 and 2.42 across 20 sessions. The latest close on file is 1.44. ACM traded between 56.41 and 63.73 across 20 sessions. The latest close on file is 57.99. SUPX traded between 6.15 and 7.09 across 10 sessions. The latest close on file is 6.72.
What the saved facts support for SuperX AI Technology (SUPX)
46 out of 100
“Anthropic, OpenAI's proposed AI risk evaluators may not have enough power to prevent disasters” scores 46 out of 100. The score uses 1 sources, and 6 linked companies.