Who Really Benefits From the Progress?

1

DTCC Launches Blockchain Securities Service

One of the world’s largest financial infrastructure providers has introduced a service for tokenising securities held in custody. Twenty financial institutions have signed access agreements, although wider adoption remains to be demonstrated.

The Depository Trust Company (DTC), a subsidiary of the Depository Trust & Clearing Corporation (DTCC), launched its Tokenization Service on 28 October, allowing eligible financial institutions to create blockchain-based representations of securities held at DTC. The digital tokens retain the ownership rights, investor protections and economic entitlements of the underlying securities. Participating institutions can convert eligible assets between conventional and tokenised forms and transfer them to supported digital wallets.

DTCC says 20 DTC participants, including Barclays, JPMorgan, Morgan Stanley and UBS, have signed agreements to access the service. It currently supports the Canton Network and DTCC’s private Besu network, with more blockchain networks planned for 2027.

The launch follows regulatory engagement, including a December 2025 no-action letter from the US Securities and Exchange Commission covering specified eligible securities. DTCC says it completed initial limited-production transactions with more than 30 firms before the launch. The company expects the service to improve collateral mobility and liquidity. Those benefits are its stated objectives, rather than independently established cost or performance improvements.

Source: DTCC

2

Opera Brings Private AI Into Its Browser

Opera’s WebAI preview allows supported AI models to operate on a user’s computer, reducing the need to send information to remote services. Website permissions and hardware requirements remain important limitations.

Opera introduced an early preview of WebAI on 27 October, bringing locally operated AI models into its Opera One browser. The technology allows websites and browser extensions to use supported AI models running on a user’s own computer. According to Opera, information processed by these models remains on the device rather than being sent to an external AI service.

Applications could include sorting browser tabs, classifying content and spotting relevant job posts. Opera says its implementation supports models including Google’s Gemma and, on compatible Macs, Apple’s on-device foundation model.

WebAI is opt-in, with models downloaded only after activation. Opera also provides site-specific permissions, allowing users to decide which websites can access the local AI capability.

The preview is available on compatible 64-bit Windows computers and Macs with Apple silicon. Opera says performance varies according to the user’s hardware, while support for Opera GX and Opera Neon is planned. Local processing offers a possible privacy advantage, although users must still understand the permissions granted to individual websites.

Source: Opera

3

Anthropic Uses Claude to Develop New Medicines

Early results from biotechnology partnerships show how AI can assist laboratory research and drug development. One experimental treatment has produced promising results in mice, but clinical effectiveness remains unproven.

Anthropic published research on 27 October describing how Claude is being used in drug development programmes involving six organisations, including Bristol Myers Squibb, BigHat Biosciences and Generate Biomedicines. The collaborations involve tasks ranging from antibody design to clinical trial planning and preparing manufacturing and regulatory documents.

At Generate Biomedicines, researchers used Claude Code to help redesign part of a potential CAR T-cell cancer therapy. According to the companies, one resulting design was approximately 70% shorter and produced roughly ten times as many receptors in laboratory tests.

Anthropic also disclosed its acquisition of Mithril Medicine, a company established by Cody Tranbarger to research treatments for two rare forms of muscular dystrophy. Tranbarger used Claude to coordinate contracts and experiments across 11 external laboratories. Anthropic reports that one programme progressed from an initial concept to promising results in mice in less than 18 months.

The company expects one candidate could be ready to enter human clinical trials as early as late 2028, provided further development succeeds. Most of the reported work remains preclinical. The research offers evidence of AI-assisted drug development, although the results do not establish that the treatments will be safe or effective in humans.

Source: Anthropic

4

MAI Image 2.7 Ranks Third. Will Designers Use It?

Microsoft has introduced an image-editing model designed to preserve details while altering scenes and objects. Strong leaderboard results suggest technical progress, but professional workflows require more than isolated editing scores.

Microsoft introduced MAI Image 2.7 on 27 October, releasing the model in preview with a focus on editing existing images rather than generating them from scratch. The model supports adjustments to lighting, individual objects, visual styles and text embedded within images. Microsoft says it can make complex changes while preserving relationships between objects and their surroundings.

According to the company, MAI Image 2.7 entered the Arena Edit leaderboard in third place and ranked second on Artificial Analysis’s image-editing leaderboard at launch. Microsoft also reports improvements over MAI Image 2.6 across all ten use cases and seven editing actions measured by Artificial Analysis. These rankings assess specific editing tasks and should not be treated as evidence of superior performance across every professional application.

The preview is available through Arena, Artificial Analysis and MAI Playground, and Microsoft says text-to-image generation will follow later. The announcement reflects growing competition around controllable AI image editing, particularly for workflows requiring changes to existing visual assets.

Source: Microsoft AI

5

Can UK Chip Investment Deliver Sovereign AI?

Britain has announced an £18 million semiconductor design academy as researchers call for greater national computing independence. The initiatives address different parts of an infrastructure challenge involving specialised hardware and access to computing resources.

The University of Edinburgh will lead a new £18 million UK Advanced Chip Design Academy, announced on 28 October as part of the government’s Autumn Budget investment package. According to the university’s announcement, the programme will provide researchers with commercial chip-design tools and opportunities to develop and test semiconductor prototypes.

Led by Professor Themis Prodromakis, the academy will operate through Edinburgh’s Centre for Electronics Frontiers and Scottish Microelectronics Centre. Its objective is to strengthen Britain’s semiconductor design expertise and connect university research more closely with hardware manufacturing.

The announcement coincided with a warning from University of Bristol supercomputing researcher Professor Simon McIntosh-Smith that the UK needs stronger domestic AI capabilities. McIntosh-Smith, director of the Bristol Centre for Supercomputing, recently gave evidence to the House of Lords Science and Technology Committee. Following that evidence, the committee selected computing capacity in the AI era as its next inquiry topic.

He identified computing infrastructure, energy availability and specialist skills as important requirements for national AI independence. Bristol operates Isambard-AI, a £225 million government-funded supercomputer that the university says supported approximately 1,000 research projects involving more than 4,000 users during its first year.

Together, the developments highlight two related priorities: developing Britain’s capacity to design advanced chips and ensuring that researchers and businesses can access the computing infrastructure needed to use them.

Source: University of Edinburgh