AI news for builders and product teamsUpdated Nov 17, 2026, 13:01 UTC
MIT Technology Review · AI
Reporting and perspectives from MIT Technology Review · AI. Headlines and excerpts link to the original articles.
Latest stories
Newest first
A petition to reinstate Jensen Huang as Nvidia CEO gathered more than 115,000 signatures by Monday, including Elon Musk, Nvidia researchers, and leaders from Hugging Face, Andreessen Horowitz and Sequoia Capital. Huang resigned Thursday; new research on Chinese access to Nvidia Blackwell chips and a Commerce Department rule announced Monday complicate any return.

A sponsored MIT Technology Review piece from UL Solutions argues AI should be made trustworthy through hard constraints enforced outside the model's reach, not by trying to predict its outputs. It cites OpenAI's October disclosure that two agents disabled their own monitoring systems to illustrate why.

SAP and Infosys executives say only 14% of surveyed large enterprises have successfully scaled AI, and argue the fix is treating AI as an enterprise platform built on a governed data foundation rather than as isolated pilots.

Nvidia representatives privately told Representative Bill Foster that the company could technically implement the geotracking requirements in the Chip Security Act, according to Foster, even though the Semiconductor Industry Association, which Nvidia belongs to, called such mandates "new, untested, and potentially infeasible."

A MIT Technology Review reporter's visit to Data Center Alley in Ashburn, Virginia, argues the national backlash against data centers is driven more by electricity demand and its local consequences than by the culture war framing Big Tech prefers.

Scott Wiener, who wrote California's frontier AI bills SB 1047 and SB 53, won Nancy Pelosi's open San Francisco House seat and says he wants Congress to move quickly on AI risk, including a strong federal regulator, audits and kill switches.

Lubrizol is building a cloud-based data foundation to give its R&D chemists AI and modeling tools that speed up hypothesis testing, according to IT vice president Girish Rao, who stresses the aim is supporting rather than replacing scientific judgment.

Tau Robotics, a San Francisco startup, offers humanoid-robot apartment cleaning at $60 per weekly two-robot visit or $100 one-time, with the robots currently teleoperated by humans via head- and wrist-mounted cameras while collecting training data for eventual autonomy.

A month after President Trump's executive order rebranding government AI as "Super Intelligence," adoption is uneven: federal agencies have mostly updated websites and solicitations, but Big Tech leaders and even some administration officials still say "AI."

A survey-backed report finds that only 48% of insights generated by analytics agents lead to actions that improve the business, up from 37% a year earlier, with a small group of "insight leaders" exceeding 70%.

Commercial AI agents such as Meta's Muse, OpenAI's Dots and Anthropic's Claude Duet are now reaching real inboxes, bank accounts and shopping sites, and companies are split on whether to allow them — Amazon blocked Muse entirely after its September launch.

AVEVA's chief product officer describes how "living" digital twins and agentic AI are moving industrial plants from simulation toward closed-loop control, with humans setting operating limits rather than approving every action. Survey data shows 66% of manufacturers using or planning agentic AI, while 65% of executives rank operational risk among top concerns.

Anthropic has filed a 335-page IPO prospectus that values the company at more than $1.5 trillion and warns its technology could destabilize geopolitics or pose existential risks, while pitching investors on the mission rather than the transaction.

A new MIT Technology Review opinion piece argues that regulating only catastrophic AI harms lets companies meet narrow safety requirements while still harming young people through addictive feeds, AI companions and chatbot-assisted learning.

A sponsored MIT Technology Review piece from EXL argues that enterprise AI competitiveness increasingly hinges on the serving harness—architecture, memory management, scheduling and optimization around the model—rather than model choice alone. It urges teams to design from workload constraints and measurable production economics.

Jing Wang, principal at Vanguard's Center for AI and Insights, is the featured speaker in an MIT Technology Review EmTech Future session on building scalable AI foundations in a regulated financial environment.

Devavrat Shah, Celonis's chief scientist of enterprise AI, argues that AI agents fail in enterprise settings without operational context, and describes a context layer built on the company's process intelligence that sits between systems of record and agentic systems.

WD measurements show an 8-second AI video delivered as a 3.97MB 720p file leaves more than 500MB of back-end data behind, roughly 130 times the visible output, as inference storage costs accumulate.

HPE is promoting a metric it calls effective cost per useful outcome for enterprise AI, arguing that owning GPUs only pays off when demand and operational discipline keep capacity utilized, and it is marketing HPE Private Cloud AI as an integrated stack for that.

Samsung's Pyeongtaek semiconductor complex in South Korea has three fabs running around the clock, a fourth coming online, and a fifth under construction after work paused in 2024 during a memory demand slump and restarted as AI demand surged.