Today’s five stories ask who decides what counts as the record, as AI reconstructs lost art history, rewrites the rules for digital dollars, joins the office, scouts dangerous ground and explains Formula 1.

1

AI recreates the lost studios of women sculptors. Can technology recover histories that were never properly recorded?

What happens when an archive contains more evidence of the obstacles artists faced than of the places where they worked? A Paris project explores that absence, with results that are both revealing and open to interpretation.

Three art historians spent a year examining letters, journals and photographs to piece together the working lives of women sculptors in nineteenth- and early twentieth-century Paris. The difficulty was not simply finding their names. Few photographs of their studios had survived.

Their research became the basis of a ten-minute documentary created by artist Prune Nourry, filmmaker Vincent Lorca and Google Arts & Culture.

As explained in Google’s 21 October announcement, the film accompanies Visionaries, Nourry’s exhibition at the Petit Palais, featuring monumental plaster portraits honouring 18 women sculptors. It runs from 20 October 2026 to 30 January 2027.

Using generative AI and image-editing tools, the team produced interpretations of workspaces associated with Nancy Elizabeth Prophet, Jane Poupelet and Camille Claudel.

A custom tool built with Gemini transformed archival photographs into three-dimensional scenes, allowing a virtual camera to move through them.

In one particularly effective sequence, women students gradually disappear from a historical photograph of Antoine Bourdelle’s studio. Only the men remain.

The intervention makes a point about how artistic reputations are constructed. Women participated in professional training and sculpture, yet their contributions were often less prominently recorded and preserved.

Nourry’s project uses technology to give those omissions a visual form.

There is, however, an important distinction between historical documentation and reconstruction. The studios are interpretations informed by research, not authenticated images of the original spaces.

Source: Google Arts & Culture

2

The US finalises stablecoin rules. Will digital dollars challenge banks or strengthen them?

Greater legal certainty could help digital payments mature. But the institutions best equipped to comply with the new requirements may also be those already dominating finance.

The Office of the Comptroller of the Currency (OCC) has issued a final rule establishing requirements for payment stablecoin issuers under its jurisdiction.

The 21 October decision implements key provisions of the GENIUS Act, covering reserve backing, redemption, custody, risk management and reporting.

The rules take effect on 18 January 2027.

Payment stablecoins are digital tokens intended to maintain a stable value against a currency, usually the US dollar. They are widely used in cryptocurrency markets and increasingly considered for international payments.

Unlike conventional bank deposits, their legal and financial arrangements vary, making the quality of reserves and redemption rights particularly important.

The OCC’s framework establishes standards for issuers under its supervision, including certain foreign companies offering stablecoins in the United States.

Anti-money laundering and sanctions requirements will be addressed separately in coordination with the Treasury.

The regulator received hundreds of comments from cryptocurrency businesses, community banks, industry associations and academics before finalising the rules.

It has not yet announced when applications from prospective issuers will begin to be processed.

Greater regulatory clarity could encourage established financial institutions to develop stablecoin services. It could also increase compliance costs for smaller companies.

Source: Office of the Comptroller of the Currency

3

Uber gives 30,000 employees AI agents. Is workplace security ready?

Connecting autonomous software to corporate accounts could change how employees work. It also creates risks that conventional chatbot policies were never designed to address.

Uber is extending ChatGPT Work to more than 30,000 employees following a pilot involving approximately 1,000 staff.

The 22 October announcement describes an agent platform capable of completing tasks across applications including Gmail, Google Drive, Google Calendar and Slack.

Employees can use it to research information, prepare meeting briefs, analyse data and produce presentations.

The distinction from a conventional chatbot is important. An AI agent can undertake a series of actions across different systems rather than simply generate a response to a question.

Uber plans to deploy the technology across operations, finance, marketing and product development, supported by approved integrations, employee training and an internal AI skills library.

Developers will continue using OpenAI’s Codex tools.

The rollout could reduce time spent on repetitive administrative work. However, access to multiple corporate systems also creates opportunities for mistakes to spread.

An agent may retrieve confidential documents, act on misleading instructions embedded in external material or attempt tasks beyond its intended permissions.

Security therefore depends on limiting access, monitoring actions and providing effective ways to interrupt or reverse operations.

Uber and OpenAI have not published independent security evaluations or detailed results demonstrating the effectiveness of the initial pilot. Uber says it will track the rollout’s impact as deployment progresses.

Source: OpenAI

4

Oxford develops robot teams for space exploration. Could they transform work on Earth?

Machines working together may be better equipped to navigate hazardous environments than individual robots. Their potential uses extend far beyond lunar missions.

Four quadruped robots and a supporting drone will form the basis of a new €1.8 million European Space Agency-funded research project involving the University of Oxford and robotics company Autodiscovery.

The initiative, announced on 21 October, aims to develop machines that share information and coordinate exploration, mapping and inspection tasks.

Known as the Threat Monitoring Networked Team, or TMNT, it will investigate how robots communicate through technologies including 5G and 6G-related systems.

Cooperative robots could cover larger areas than individual machines and continue operating when one encounters an obstacle or technical problem.

Researchers plan to test the system at the ESA’s LUNA facility, which simulates aspects of the lunar environment, and at sites such as factories, university campuses and data centres.

Potential applications include inspecting power stations, monitoring critical infrastructure and exploring hazardous terrain.

The same capabilities could support scientific missions or reduce the need for workers to enter dangerous locations.

But coordinating several autonomous machines remains technically difficult.

Robots must combine information from multiple sensors, navigate changing environments and maintain reliable communications without interfering with one another.

The project will investigate these challenges through field trials. It is not yet an operational system proven across the environments proposed.

Source: University of Oxford

5

Ferrari brings AI closer to Formula 1 fans. Does it make the sport more accessible?

Explaining race strategy is a practical use for AI. Whether it improves the experience is a different question from whether it keeps supporters using an application longer.

A new Race Timeline in Ferrari’s mobile application allows supporters to follow pit stops, fastest laps, changing weather conditions and other developments as a Formula 1 session unfolds.

The feature forms part of an IBM and Ferrari update announced on 22 October.

AI agents built with IBM’s watsonx technology also produce summaries of practice sessions, qualifying and races, using performance data and information about the team’s strategy.

IBM says the reports can be generated within minutes of each session ending.

Other additions include a pit-stop game, predictions, quizzes and a section where supporters can share six-word stories about Ferrari.

The companies want to make racing information easier to understand while encouraging interaction with the team.

IBM also reports improvements in the application’s development and usage.

It says more than 70 reusable AI skills and more than 30 agents contributed to a 30% overall productivity improvement in the app’s product development lifecycle.

Following the application’s May 2025 relaunch, IBM recorded a 35% increase in cumulative downloads and a 36% rise in average monthly active users during the first year.

These are company-reported measurements. They do not establish how much of the improvement resulted directly from AI or whether users found the experience more satisfying.

The features nevertheless offer a concrete application of generative AI: organising detailed information and explaining complex events to a wider audience.

Source: IBM