Ten stories moved AI this week. Money led the week. Two labs landed under their own asking price, and a federal task force opened its doors.
Its law dates to 1986. An outside auditor said a frontier lab now moves twice as fast, then blacked out most of the proof.
Justice Department Opens Its AI Task Force
Attorney General Todd Blanche named the body Thursday. He called it the Super Intelligence Task Force. Stanley Woodward and Stephen Weiss will co-lead it1.
Its mandate spans foreign state misuse, organized fraud and corporate crime, and it folds in a January unit too.
Weiss described what companies should expect. "The Department also has a responsibility to utilize law enforcement tools at our disposal to deter improper use of SI, harden regular reporting on SI risk and response, and ensure companies exercise vigorous internal controls," he said1.
The release skips naming any new crime of its own. It leans entirely on laws already on the books. Georgetown law professor Paul Ohm told senators in September that the task force's own test case may miss the mark.
OpenAI's agents broke into Hugging Face in July, acting on their own, apart from a person's direction2. The hacking statute at issue dates to 1986, before autonomous software existed. A Sept. 29 executive order gives the White House science adviser until Nov. 28 to propose a legal definition of the term.
Former FTC chair Lina Khan reads the gap differently. Regulators can already charge companies and their bosses over unsafe products, she posted. Her theory rests on product law, a path apart from the hacking statute Ohm doubts15.
Lina Khan@linamkhanLaw enforcers already have authority to charge companies and their CEOs for creating and releasing dangerous, unvetted, or defective products. We shouldn’t let discussions about new legal regimes distract from the fact that there’s no AI exemption from laws already on the books —…
OpenAI Closes $11 Billion, Short of Its Ask
A new round landed Friday. OpenAI wired $11 billion, against a target that once ran as high as $30 billion3. The close values the company at $1.25 trillion.
That sits below the $1.4 trillion OpenAI sought before a selloff in AI stocks. SoftBank and MGX joined the round. It could still grow toward $20 billion in coming weeks.
Yearly sales sit near this story's center. OpenAI told investors its sales run close to $50 billion a year. That is $20 billion below an earlier guess, one built by comparing OpenAI's numbers to Anthropic's own method4.
Gene Munster, managing partner at Deepwater Asset Management, reads the swings as noise over signal. "It can miss the bigger picture. The bigger picture: it's still growing, we're still early in AI, and they're still in a great place," he told Fortune5.
OpenAI's Price Settles at $1.25 Trillion, Anthropic's Low End Sits at $1 Trillion
The valuation each company once sought against this week's price, in trillions of dollars
The blue dot is the value each company once sought; the magenta dot is this week's price; the rod spans the drop between them.
OpenAI's $1.25 trillion is a closed round. Anthropic has yet to price its shares; $1 trillion is only the low end of the range investors discussed this week.
Sources: Reuters [3]; CNBC [6]. Chart by Ryan Elliott Dennis.
The numbers behind this chart
| Item | Original target | This week's price |
|---|---|---|
| OpenAI | $1.40T | $1.25T |
| Anthropic | $2.00T | $1.00T |
Anthropic's Bankers Hear the Word Stretch
Its own IPO math moved this week too. Three big investors told CNBC that Anthropic's price will likely land near $1 trillion to $1.3 trillion. That sits below the $2 trillion the company's own filing once sought6.
The Nasdaq fell more than 5 percent Thursday, its steepest drop in 19 months. A Democratic sweep in the midterms drove the fall, on fears of tighter rules on data centers.
Gil Luria, an analyst at D.A. Davidson, frames the lower number as the surer path. "The market really needs this IPO to go well and the bankers on the deal surely understand that. The best way to set Anthropic up for success is to price the deal well," he said6.
Aswath Damodaran, the NYU valuation professor who publishes his own Anthropic model, flags the assumption the whole price rests on. "The fragility of the target revenue assumption comes from it being susceptible to any of the parts failing, and if a part fails... that, in turn, will have ripple effects on the other parts of your story," he wrote7.
METR Finds Anthropic Building Twice as Fast, Redacted
The nonprofit that evaluates frontier labs, METR, spent 11 weeks inside Anthropic's own data. Its finding: Anthropic's pace of internal research rose to 1.9 times its earlier rate after October 2025. METR's 95 percent confidence range runs from 1.5 to 2.88.
Roughly 80 percent of that speedup traces to AI helping human researchers, METR guesses. Faster chips play a smaller part. Still, the reviewers struggled to check their own toughest findings.
Anthropic redacted most of the evidence METR called most useful. Outsiders must take the summary on trust, apart from the record behind it.
Anthropic's Research Pace Rises to 1.9 Times Its Old Rate, METR Finds
METR's measure of Anthropic's internal research speed, before and after October 2025
Each column is Anthropic's research pace relative to its rate before October 2025; the taller column shows research now moving 1.9 times as fast.
METR's 95 percent range runs from 1.5 to 2.8, and it says Anthropic redacted most of the evidence it called most useful.
Source: METR [8]. Chart by Ryan Elliott Dennis.
The numbers behind this chart
| Item | Value |
|---|---|
| Before Oct. 2025 | 1x |
| After Oct. 2025 | 1.9x |
Mistral Puts Managed Agents to Work Inside Companies
A French lab moved into a new business this week. Mistral expanded its Studio platform Nov. 5 with managed agents9. The company now runs this software on a customer's own systems, on the customer's behalf.
Each agent can follow a saved procedure, run sandboxed code and hand its task to a second agent.
Enterprise apps arrive alongside the agents. Each one carries its own hosting, authentication and access controls. Private cloud and on-premises versions reach customers by the end of November, timed to Mistral's newest large model.
Adobe's Narayen Steps Down Dec. 1
Shantanu Narayen will hand Adobe's chief executive title to Anil Chakravarthy on Dec. 1. The move ends an 18-year run atop the company10. Narayen stays on as board chairman through the transition.
Investors pushed for the change. Adobe's stock sits down 25 percent this year. Shareholders read the company's AI revenue as too small for the price it still commands10. Chakravarthy currently runs Adobe's customer-experience business and its worldwide field operations.
Hollywood's Agents Back the Times Against OpenAI
WME filed a brief Oct. 30. It backs the New York Times in its copyright suit against OpenAI and Microsoft11. The talent agency joined CAA, the Human Consent Fund, SAG-AFTRA and the Writers Guild. All four argue one point: training a model on copyrighted work still calls for a license.
Reply briefs in the case fall due Nov. 20. Both sides already filed competing summary-judgment motions this fall11.
OpenAI Hires Anthropic's Claude Code Lead
Catherine Wu announced her Anthropic exit Nov. 3. She landed at OpenAI two days later, where she will run product for the company's core tools12. Wu had led product for Claude Code, Anthropic's coding agent, since joining the company in 2024.
Her move makes three departures in two weeks12. Before Anthropic, Wu worked as a product engineer at Scale AI and Dagster, then spent a stretch as a venture capitalist.
PitchBook Prices Databricks at a Third
PitchBook's own research team values Databricks near $68.7 billion. That marks a 64 percent discount to the $190 billion price its August funding round set13. Databricks raised that round on a $7 billion annualized run rate, 27 times revenue.
Even PitchBook's most generous scenario tops out near $182 billion. That scenario still assumes an 80 percent gross margin Databricks has yet to reach13. Today's planned Money & Finance column weighs that same gap: does $190 billion still fit a software company growing at this pace?
October's Jobs Report Tests the Hiring Freeze
Entry-level job postings fell 7.5 percent over the past year. Postings for senior roles climbed 14.7 percent over that same stretch14. Today's planned Talent column reads the October jobs report against that split. Its question: does official data now show the AI hiring freeze reporters have described all year?
What to watch
Watch Nov. 28. Blanche's executive order sets that date as the deadline for a formal legal definition of the term super intelligence.
Track Anthropic's actual roadshow price against the $1 trillion floor this week's doubters named. A real number beats any guess. METR's next review deserves a second look too. A wider release of its underlying evidence would answer the redaction question this week's report left open.
Sources
- "Department of Justice Launches Super Intelligence Task Force," U.S. Department of Justice, Office of Public Affairs, Nov. 5, 2026, https://www.justice.gov/opa/pr/department-justice-launches-super-intelligence-task-force
- Mark Rutherford, "DOJ Super Intelligence Task Force Launches With No New Offense; Hacking Law Needs Human Intent," Tech Times, Nov. 6, 2026, https://www.techtimes.com/articles/330168/20261106/doj-super-intelligence-task-force-launches-no-new-offense-hacking-law-needs-human-intent.htm
- "OpenAI gets $11 billion funding at $1.25 trillion valuation, source says," Reuters, Nov. 6, 2026, https://lufkindailynews.com/news_reuters/business/openai-gets-11-billion-funding-at-1-25-trillion-valuation-source-says/article_030d6022-41a7-5362-983c-56f015019b13.html
- Lucas Ropek, "OpenAI's revenue is reportedly $20 billion less than previously projected," TechCrunch, Oct. 8, 2026, https://techcrunch.com/2026/10/08/openais-revenue-is-reportedly-20-billion-less-than-previously-projected/
- "Wall Street is panicking about OpenAI. A veteran tech analyst says everyone's overreacting," Fortune, April 28, 2026, https://fortune.com/2026/04/28/markets-openai-revenue-oracle-softbank-coreweave-sam-altman-trial-elon-musk/
- Kate Rooney and Caleigh Keating, "Investors brace for lower Anthropic valuation, calling $1.5 trillion a 'stretch'," CNBC, Nov. 6, 2026, https://www.cnbc.com/2026/11/06/investors-brace-for-anthropic-valuation-1point5-trillion-a-stretch.html
- "'Dean Of Valuation' Aswath Damodaran Trims His Anthropic IPO View, Flags 'Fragility' Of $1-Trillion Revenue Target," Yahoo Finance, Nov. 4, 2026, https://finance.yahoo.com/markets/stocks/articles/dean-valuation-aswath-damodaran-trims-090204116.html
- "Review of the 'Risks from automated R&D' section in Anthropic's Risk Report (August 2026)," METR, Nov. 5, 2026, https://metr.org/blog/2026-11-05-rd-section-anthropic-risk-report-august-2026-review/
- "Build Enterprise Agents, Workflows & Apps in Mistral Studio," Mistral AI, Nov. 5, 2026, https://mistral.ai/news/build-with-studio/
- "Adobe CEO Shantanu Narayen to Step Down After Successor Named," TechBuzz.ai, Nov. 4, 2026, https://www.techbuzz.ai/articles/adobe-ceo-shantanu-narayen-to-step-down-after-successor-named
- "WME Joins New York Times Copyright Battle Against OpenAI," Deadline, Nov. 2, 2026, https://deadline.com/2026/11/openai-lawsuit-copyright-wme-nwe-york-times-ai-1237211006/
- "Anthropic loses product lead Catherine Wu behind Claude Code to OpenAI," The Decoder, Nov. 5, 2026, https://the-decoder.com/anthropic-loses-product-lead-catherine-wu-behind-claude-code-to-openai/
- "The math behind Databricks' $190B price doesn't add up, yet," PitchBook, Nov. 5, 2026, https://pitchbook.com/news/articles/databricks-190-billion-valuation-overvalued-research
- "Entry-Level Hiring 2026: Postings Down 7.5%, Junior Reqs Hold," Pin, Nov. 5, 2026, https://www.pin.com/blog/entry-level-hiring-collapse-2026/
- Lina Khan, post on X, Sept. 13, 2026, https://x.com/linamkhan/status/2099204390548639960
