This Week's Top Five Stories in AI

Why Uber has Already Burned Through its AI Budget

All good things come at a cost and, for Uber, that is the price of its heavy AI adoption.

The transportation tech company has already exhausted its AI budget for 2026 due to surging use of coding tools, according to the company’s CTO Praveen Neppalli Naga. 

Uber’s total research and development (R&D) expenses went up 9% year-over-year in 2025, hitting US$3.4bn, with AI a key cost driver. 

As reported by The Information, engineers are extensively using tools like Claude Code, with about 11% of Uber's live backend code updates being completely written by AI agents. 

Praveen says Uber teams are heading “back to the drawing board” as it counts the costs of surging AI usage.

How Bloomberg is Using Agentic AI for Complex Workflows

With financial market decisions capable of moving billions in seconds, there is zero tolerance for any errors. Bloomberg’s latest update, announced at its AI in Finance Summit, optimises the investment process with the help of agentic AI.

The development is an evolution of Bloomberg’s existing conversational AI interface ASKB that redefines how financial professionals discover, analyse and act on information using the Bloomberg Terminal. 

Used by decision-makers for more than four decades, Terminal provides real-time access to news, data, insights and trading tools that help customers turn knowledge into investment decisions.

ASKB will transition from a simple search/discovery tool into an integrated engine for “institutional intelligence”, providing investors with the information they need to accelerate their work through a coordinated network of Bloomberg’s AI agents.

Wayne Barlow, Global Head of Terminal Products at Bloomberg, says: “We are providing investors with an agentic AI engine that synthesises a comprehensive web of interconnected market data with a firm’s unique proprietary knowledge so they can hone their unique, alpha-generating perspectives.”

GPT 5.4-Cyber & OpenAI's Trusted Access for Cyber Programme

There is new wind in the arena for AI cyber defence and it is coming from OpenAI’s new release: GPT‑5.4‑Cyber. 

Committing US$10m in API credits through its Cybersecurity Grant Program and scaling its Trusted Access for Cyber (TAC) to defenders, OpenAI is strengthening its position among cyber defenders. 

OpenAI expects cybersecurity to become increasingly proactive rather than reactive, with intelligent systems playing a central role in detecting and preventing threats before they escalate. 

As AI cyber capabilities can power both offence and defence, OpenAI's TAC ensures that access to such powerful tools rests firmly in the hands of the defenders. 

Jeff Bezos' Project Prometheus Raises US$10bn

Investor confidence in AI shows no sign of slowing down, with Jeff Bezos’ Project Prometheus moving to raise US$10bn in its latest funding round. 

Following this, the AI lab will now be valued at a whopping US$38bn, a Financial Times (FT) report notes.  

The funding was backed by JPMorgan and BlackRock, according to industry reports. 

Co-founded by Jeff Bezos back in November 2025, Project Prometheus is shrouded in secrecy, with a mostly blank LinkedIn page and a description that reads "AI for the physical economy".

It is run by Co-CEO and Co-founder Vik Bajaj, who previously served as a director at Google’s moonshot factory, X. 

Mythos AI: Anthropic Investigates Unauthorised Access Claims

With the capabilities of Anthropic’s Claude Mythos Preview already well documented, recent reports of possible "unauthorised access" to the model will unsettle many. 

“A handful of users in a private online forum gained access to Mythos on the same day that Anthropic first announced a plan to release the model to a limited number of companies for testing purposes,” states a Bloomberg report. 

Mythos – Anthropic’s most powerful model yet – was able to find thousands of vulnerabilities in everyday software and is capable of chaining together these bugs to create complicated attack chains that can compromise these platforms. 

For this reason, the company had kept the model from the public, releasing it only to major industry players as part of a security coalition that aims to secure critical software, named Project Glasswing. 

The coalition includes AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA and Palo Alto Networks.