Current affairs
What the builders are saying
Public statements, corporate documents, court testimony and internal records from the companies that design digital environments and artificial intelligence systems.
Each entry reproduces the exact words of its author, states where and when they were said, and links to the source. A new edition every Monday.
Nature of the documents
- Public statement. Said in public by the person to whom it is attributed.
- Corporate document. Published by the company itself through an official channel.
- Disclosed internal document. A document not meant for publication, brought to light in legal proceedings.
- Court filing or testimony. A court filing or a statement given before a court.
- Insider account. An account by someone who worked inside the organisation.
Edition I · 14 September 2026
This first instalment sets the baseline: the statements, corporate documents and court records of the past year that define the industry’s public position. On 12 September the chief executive of Anthropic called for slowing the pace at which models improve, three days after a researcher at his own company resigned accusing the sector of the opposite. In parallel, the federal trial in Oakland brought to light internal Meta documents on how its products for minors were designed.
-
Dario Amodei12 Sep 2026
“We must slow the pace at which we improve the capabilities of AI models.”
co-founder and chief executive
AnthropicAI laboratories · Corporate document
The head of one of the three frontier laboratories states in writing that the current pace of capability improvement is excessive. It sets a declared standard against which his own company’s deployment calendar can later be measured.
Essay “We Must Pace the Frontier”, on his personal site · 12 September 2026
-
Dario Amodei12 Sep 2026
“Not building the technology deprives humanity of benefits or simply places AI in the hands of authoritarian powers, while building it too fast is reckless.”
co-founder and chief executive
AnthropicAI laboratories · Corporate document
This is the canonical formulation of the argument by which frontier laboratories justify continued deployment while acknowledging serious risk. It documents the structure of the reasoning, not only its conclusion.
Essay “We Must Pace the Frontier”, on his personal site · 12 September 2026
-
Jacob Coxon9 Sep 2026
“Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives.”
pre-training researcher, resigned
Anthropic and OpenAIAI laboratories · Insider account
Testimony from a researcher with direct access to pre-training at both companies, three days before Amodei’s essay. He forfeited unvested equity, which removes financial incentive as an explanation for his earlier silence.
Own statement on X, reported by TIME · 9 September 2026
Source
The original is a post on X. The link goes to the TIME interview, which reproduces the text.
https://time.com/article/2026/09/09/ai-anthropic-openai-jacob-coxon/
-
Demis Hassabis14 Jul 2026
“We’ve essentially found a way to make sand think. It’s miraculous.”
co-founder and chief executive
Google DeepMindAI laboratories · Public statement
In the same piece he places artificial general intelligence a few years away and calls for an international watchdog before year’s end. It documents a chief executive requesting the external control his own company does not impose on itself.
Own manifesto, reported by Axios · 14 July 2026
-
Dario Amodei26 Jan 2026
“AI is increasingly matching the general cognitive profile of humans.”
co-founder and chief executive
AnthropicAI laboratories · Corporate document
A claim by the manufacturer itself that its systems match the general human cognitive profile, not isolated capabilities. It is the premise on which arguments of substitution and cognitive delegation subsequently rest.
Essay “The Adolescence of Technology”, on his personal site · 26 January 2026
Source
The essay prints only “January 2026”; the day comes from Axios and Fortune coverage.
-
Anthropic18 Dec 2025
“Sycophancy means telling someone what they want to hear—making them feel good in the moment—rather than what’s really true.”
corporate communications
AnthropicAI laboratories · Corporate document
A company in the sector defines sycophancy as a design failure with consequences for the user. It is a primary admission that tuning for stated preference can degrade the truthfulness of the system.
“Protecting the wellbeing of our users”, corporate blog · 18 December 2025
-
OpenAI25 Nov 2025
“his death, while devastating, was not caused by ChatGPT”
answer filed in Raine v. OpenAI
OpenAIAI laboratories · Court filing or testimony
This is the company’s formal legal position on psychological harm attributed to its product, in the case of the suicide of a sixteen-year-old. The defence shifts causation to the user and pleads breach of the terms of service.
California Superior Court, reported by Bloomberg · 25 November 2025
-
Sam Altman14 Oct 2025
“as part of our “treat adult users like adults” principle, we will allow even more, like erotica for verified adults”
co-founder and chief executive
OpenAIAI laboratories · Public statement
The relaxation of content restrictions is announced on the claim that serious mental health problems have already been mitigated, without publishing the evidence for it. The feature remained unreleased five months past the announced date.
Post on X, reported by TechCrunch · 14 October 2025
Source
The original is a post on X. The link goes to the report that reproduces the text.
https://techcrunch.com/2025/10/14/sam-altman-says-chatgpt-will-soon-allow-erotica-for-adult-users/
-
Sam Altman16 Sep 2025
“We prioritize safety ahead of privacy and freedom for teens; this is a new and powerful technology, and we believe minors need significant protection.”
co-founder and chief executive
OpenAIAI laboratories · Corporate document
The company acknowledges in writing that its own guiding principles conflict, and sets an order of precedence it decides unilaterally. It is primary material for examining who currently sets the thresholds of cognitive protection for minors.
“Teen safety, freedom, and privacy”, OpenAI corporate blog · 16 September 2025
-
Arturo Béjar19 Aug 2026
“Most people will not turn the setting on. It’s like you have to turn on the air bag every time you get into the car.”
former safety engineer, testimony under oath
formerly Meta PlatformsPlatforms and networks · Court filing or testimony
A former internal lead testifies that real protection is determined by default settings, not by available options. He further held that safety was not a meaningful priority inside the company.
Federal trial in Oakland, reported by NPR · 19 August 2026
-
Meta Platforms18 Aug 2026
“Long Term Retention: The Young Ones Are The Best Ones”
internal retention study, disclosed at trial
Meta PlatformsPlatforms and networks · Disclosed internal document
The title of an internal study that treats early acquisition as a long-term retention asset. The same session cited a 2016 email setting teen time spent as the overall company goal for Instagram.
Federal trial in Oakland, reported by NPR · 18 August 2026
-
Meta researcher30 Jan 2026
“oh my gosh yall IG is a drug. We’re basically pushers… We are causing Reward Deficit Disorder”
user experience · internal message from 2020, disclosed in litigation
Meta PlatformsPlatforms and networks · Disclosed internal document
Internal acknowledgement, by the company’s own technical staff, of a compulsive reinforcement mechanism in the product. It is a discovery exhibit, not a public position of Meta.
Court exhibit reproduced by The Markup · 30 January 2026
-
Mark Zuckerberg30 Jan 2026
“If we tell teens’ parents about their live videos, that will probably ruin the product from the start.”
internal email from 2016, disclosed in litigation
Meta PlatformsPlatforms and networks · Disclosed internal document
On teenagers’ live video, not informing responsible adults appears treated as a condition of the product’s viability. Alerting parents is weighed by its effect on usage, not by its effect on the minor.
Court exhibit reproduced by The Markup · 30 January 2026
-
Karandeep Anand11 Nov 2025
“the longitudinal impact of chatbot interaction could be unhealthy, or is not fully understood”
chief executive
Character.AIPlatforms and networks · Public statement
The company admits uncertainty about the long-term effects of conversational companions as it bars open-ended conversation for under-eighteens. The chief executive himself describes the measure as reversible.
Interview with TIME · 11 November 2025
-
Elon Musk19 Sep 2025
“The algorithm will be purely AI by November, with significant progress along the way.”
owner
X Corp.Platforms and networks · Public statement
The platform’s owner announces replacing heuristic curation with an artificial intelligence model under his control. The change was confirmed operational in November 2025.
Post on X, reported by Social Samosa · 19 September 2025
Source
The original is a post on X. The link goes to the report that reproduces the text.
https://www.socialsamosa.com/news-2/x-ai-powered-algorithm-10486616
-
Mark Zuckerberg30 Jul 2025
“if you don’t have glasses that have AI […] you’re probably going to be at a pretty significant cognitive disadvantage compared to other people”
chief executive
Meta PlatformsPlatforms and networks · Public statement
A company product is presented as a condition of cognitive competition between people. It is an explicit formulation of dependence on a device carrying permanent sensors over the user and their surroundings.
Second-quarter earnings call, reported by TechCrunch · 30 July 2025
-
Alex Karp13 Nov 2025
“Not only was the patriotism right, the patriotism will make you rich.”
co-founder and chief executive
Palantir TechnologiesSurveillance and data · Public statement
It explicitly links alignment with the state to financial return, and answers objections about surveillance by shifting the debate to stock performance.
Yahoo Finance Invest Conference, reported by Fortune · 13 November 2025
-
Alex Karp6 Nov 2025
“We are monetizing the fact that these decisions are difficult.”
co-founder and chief executive
Palantir TechnologiesSurveillance and data · Public statement
The chief executive describes the business model as monetising the trade-off between security and civil liberties. In the same interview he attributes ninety-eight per cent of everyday monitoring of citizens to private commerce.
Video interview with Axios · 6 November 2025
-
Elon Musk31 Dec 2025
“Neuralink will start high-volume production of brain-computer interface devices and move to a streamlined, almost entirely automated surgical procedure in 2026.”
founder
NeuralinkNeurotechnology · Public statement
An announcement of industrial scaling of brain implants by its founder. It sets a horizon of mass deployment while legal protection of neural data exists in barely five states of a single country.
Post on X, reported by Fierce Biotech · 31 December 2025
Source
The original is a post on X. The link goes to the report that reproduces the text.
No entry matches those filters.
A note on method
Only words verified against the linked source are carried here: the original document where it is reachable and, where it is not, the outlet that reproduces it, stated in each entry. Quotes are published in their original language. Internal records obtained through litigation are identified as such and do not amount to a public position of the company. To report an error or a wrong attribution, write to the Observatory.