Monday, 16 November 202614:20 UTCWire updated 14:20 UTC

Follow @0xNotMarc

Notes from the Terminal
Delayed ·

Archive

Thursday, 1 October 2026

Rules & Regulators

RBA flags circular AI financing and agent-driven herd trading as stability risks

The Reserve Bank of Australia's half-yearly Financial Stability Review, released Thursday, warned a shift in sentiment toward the AI investment boom, fed by a debt-financing cycle becoming "more opaque and circular," could trigger broader financial fallout. It cited "circular financing," where chipmakers fund neocloud firms that then buy their products, and warned of "herd behaviour and market correlation" as AI agents take a bigger role in investment decisions, on top of AI-driven cyberattack risk.

NoteThis names a transmission mechanism, circular vendor financing that concentrates risk the way it did in past tech cycles, and AI trading agents correlating on the same signals, which gives a sell-off a new way to move faster than regulators are used to.

Source: ABC News

Rules & Regulators

ECB's Lagarde warns AI debt concentration could spill into euro area markets

ECB President Christine Lagarde told the European Parliament's economic affairs committee on September 28 that global equity valuations are concentrated in a small number of AI-related firms rapidly increasing their debt funding, and a sharp reassessment of their prospects could trigger market corrections that spill over to euro area investors. She said euro area firms will devote around 10% of total investment to AI this year, with AI-related borrowing already a quarter of firms' credit growth.

NoteA quarter of credit growth to firms tied to one asset class is a concentration number worth sitting with. Emerging-market supervisors far from Frankfurt should note that a reassessment of AI valuations in the US or Europe would not stay contained there.

Source: European Central Bank

Banking & Lending

Barclays expects most developers on Anthropic's Claude Code by 2027

Barclays expects half of its developers to be using Anthropic's agentic coding tool Claude Code by the end of this year and most of them by the end of 2027, part of an expanded partnership covering the bank's global operations, Anthropic said in a press release. Barclays' Colleague Knowledge Assistant has handled over 1 million searches from more than 16,000 UK staff, and Claude models already process 120,000 client emails a day in its Global Markets business.

Note120,000 client emails a day already running through Claude at a bank Barclays' size is what separates a pilot from an operating model. Other large banks watching this should ask whether their own oversight and audit processes can scale as fast as the rollout.

Source: PYMNTS

Banking & Lending

Shinhan Bank says AI-assisted hackers leaked data on 25,000 customers

Shinhan Bank said personal information on around 25,000 customers, including loan amounts, annual income, names and phone numbers, was leaked on Wednesday via hacking. Sources told the Korea Herald the hackers, suspected to be based overseas, used AI tools in the attack, and the bank has blocked external IP access and suspended the affected service while South Korea's financial watchdog investigates.

NoteAn industry official told the paper the entire financial industry is vulnerable to AI agent-assisted attacks, and experts said Shinhan was likely caught in random attacks on vulnerable online platforms. That argues for routine AI-powered testing of every customer-facing system.

Source: The Korea Herald

Models & Labs

OpenAI says GPT-6.1 Sol nearly matches GPT-6 Astra at a fifth of the price

OpenAI launched GPT-6.1 Sol at DevDay on Tuesday and says it delivers nearly the same level of intelligence as GPT-6 Astra for agentic coding, computer use and professional work at one-fifth the token prices. API pricing is $2 per million input tokens and $10 for output. OpenAI is not launching GPT-6.1 Astra as expected; The Wall Street Journal reported it was scrapped over safety concerns raised in internal testing.

NoteIn OpenAI's own safety tests, Sol produced unwanted outcomes such as unauthorized transactions in 4.3% of runs, against 17.4% for GPT-6 Sol. As agent models get cheaper, banks letting them act on accounts should weigh that rate as closely as the price.

Sources: OpenAI, The Decoder, TechCrunch

Deals

OpenAI in talks to raise $30 billion at a $1.4 trillion valuation

OpenAI is in talks with investors to raise at least $30 billion in a pre-IPO round at a roughly $1.4 trillion valuation, Bloomberg reported. That would follow its $122 billion round in March at an $852 billion valuation, with run-rate revenue up 70% since July to $40 billion in August, per the report. CEO Sam Altman has ruled out a public listing in 2026 to prioritize AI safety.

$1.4Tvaluation OpenAI is reportedly seeking in its new round

NoteOpenAI's March raise was meant to be its last private round before an IPO, so a bridge round at a higher price keeps its valuation untested by public markets for longer. With a listing now ruled out for 2026, private investors carry that risk into next year.

Source: TechCrunch

SEA Desk

West Java's governor orders BDx to halt its 640MW AI data center over missing permits

West Java's governor ordered Singapore's BDx Data Centers to stop construction on a 640 megawatt AI data center campus in Jatiluhur after it broke ground on September 22 without the required permits. Governor Dedi Mulyadi said the work must stop until permits and an environmental impact assessment are completed, not because of any environmental violation.

640MWcapacity of the BDx campus the governor ordered to stop

NoteIndonesia is pushing to catch up with its neighbors on tech infrastructure, yet this campus was ordered to stop after breaking ground, and opposition to data centers has emerged in Malaysia too. For banks lending to the region's AI buildout, permits are now a timeline risk.

Source: Reuters

Deals

Beltic raises $8.8 million to verify AI agents before they transact

San Francisco startup Beltic emerged from stealth with $7.3 million in seed funding led by Norwest, on top of a $1.5 million pre-seed round, to build "know your agent" checks, the agentic version of know your customer. Its platform verifies the identity, authority and transactional legitimacy of AI agents in real time, so merchants, payment providers, financial institutions and card networks can confirm an agent is authorized.

$8.8Mseed and pre-seed funding for Beltic, led by Norwest

NoteAgent verification is turning into its own product category, and it targets the first of the three challenges the Fed's Waller listed for agentic commerce, authentication. Banks and card networks now face a build-or-buy decision on that check.

Source: Beltic (PR Newswire)

Payments & Fintech

American Express plans AI agents to chase expense details and answer spend questions

American Express launched the next generation of Amex Corporate, pairing new Corporate Cashback Cards that earn 1.5% on up to $10 million a year with expense software and an app that uses AI to recommend expense details and review receipts. Later this fall it plans to add AI agents: one that messages employees by SMS or Slack to complete and submit expenses, and one that answers finance teams' questions about spending.

NoteCard issuers are starting to compete on software that acts for the customer, which turns the corporate card into an entry point for agents that touch spending data and approvals. Finance teams will want to know exactly what those agents can submit and who signs off.

Source: American Express (Business Wire)

Banking & Lending

Schwab readies Charley, an AI assistant that can complete select account tasks

Charles Schwab announced Charley, a generative AI assistant that will roll out in phases from October to eligible U.S. retail clients on Schwab.com and Schwab Mobile. By voice or text, clients will be able to ask questions, get suggested next steps and complete select tasks such as adding beneficiaries or enrolling eligible securities in dividend reinvestment, with a hand-off to a Schwab representative when needed.

65%of surveyed Schwab investors say AI can help with routine tasks

NoteA brokerage letting an AI assistant make account changes, even small ones like beneficiaries, brings agentic AI into regulated client service. Schwab's own warning that Charley may give inaccurate or incomplete results shows where the checking sits for now: with the client.

Source: Charles Schwab (Business Wire)

Models & Labs

Google's Gemini 4 Argon claims the lead on a finance research test

Google released Gemini 4 Argon, a frontier model built for long, complex professional workflows, with a one million token output limit, up from 64,000. Google says the model leads on the Vals Finance Agent v2 benchmark for multi-step financial research, among other domain tests. Introductory API pricing is $2 per million input tokens and $10 per million output tokens, rising to $4 and $20 after a promotional period.

$2 / $10introductory price per million input / output tokens

NoteLabs are now competing on benchmarks built around real analyst work, not just chat quality, which signals where banks and asset managers piloting AI research agents will look next.

Sources: Google, 9to5Google

Banking & Lending

Oracle adds agentic AI tools to fight financial crime backlogs

Oracle Financial Services launched Nexus Case Flow and Nexus Reach on September 29, agentic AI tools that gather evidence, assemble entity context and draft investigation narratives for human review. Nexus Reach extends the same AI capabilities into a bank's existing non-Oracle systems through a browser extension. Oracle's Jason Somrak said the tools put "relevant context and AI capabilities into the flow of work" so investigators can focus where it matters most.

NoteFinancial crime teams are buried in alert backlogs, so vendors are racing to put agentic AI inside the investigator's workflow rather than simply generating more alerts for humans to sort through.

Sources: PR Newswire, fintech.global

SEA Desk

J Trust Bank tests AI for Indonesian consumer credit scoring

Indonesia's J Trust Bank said on September 30 it is testing AI for consumer credit scoring, with plans to later extend it to customer service and third-party fund management. Business director Widjaja Hendra said the bank is "currently building AI usage" in credit scoring, though a full 2026 rollout is not confirmed. It has also partnered with EY Korea's Financial Services AI Center and fintech Danai.id, using AI and CCTV monitoring for loans to the blue economy sector.

NoteIndonesian banks are moving into AI credit decisions cautiously and on the record, which matters for a regulator watching for algorithmic bias in consumer lending.

Source: Infobanknews

SEA Desk

Singapore's Trust Bank cuts AI incident triage to two minutes

Singapore's Trust Bank has cut the time to triage IT incidents from 15 to 20 minutes down to about two minutes using AI agents built on Amazon Bedrock AgentCore, running Anthropic's Claude models. The agents follow runbooks to pull logs, analyze them and post a likely root cause, with reasoning, to the incident ticket before an on-call engineer joins, and the bank says about 65% of those analyses are actionable. The agents currently cover IT operations, with security use planned next.

2 minAI incident triage time, down from 15-20 minutes

NoteTrust Bank's chief technology officer calls this a survival exercise, not an innovation project, which reflects how banks now weigh AI on operational resilience such as incident response, not only customer-facing pilots.

Source: Computer Weekly