OpenAI Pushes Deeper Into Finance
Plus: Regulators convene bankers on Anthropic's Mythos risk
Hey, it’s Matt. You’re reading AI Street, where I report on how Wall Street uses AI.
Manage Email Preferences
If you prefer to receive one weekly email with all AI Street content, uncheck Research and Interviews.
NEWS
Anthropic’s Mythos Triggers Bank Cyber Risk Talks
Anthropic’s new model, Mythos, is being treated as a potential systemic risk to the financial system.
Anthropic says Mythos identified and exploited vulnerabilities in real-world software, including bugs that had remained unpatched for 10 to 20+ years, with one example dating back 27 years. Treasury Secretary Scott Bessent and Fed Chair Jerome Powell convened Wall Street executives to assess whether the model could increase cyber risk across financial institutions.
From Bloomberg:
The previously unreported meeting, arranged on short notice, is another sign that regulators consider the possibility of a new breed of cyber attacks as one of the biggest risks facing the financial industry. All the banks summoned to the meeting are classified as systemically important by top regulators, meaning their stability is a priority for the global financial system.
Beyond the immediate cyber concern, the episode is another sign that model capabilities are still improving.
A year to 18 months ago, there were concerns that LLMs were approaching the limits of scaling, with diminishing returns from additional data and compute.
Models have continued to improve, with gains increasingly coming from post-training and system design rather than larger pre-training runs.
The same architecture is being adapted to other data types. Netflix and Stripe have both reported improvements after training models on internal datasets.
Related:
Mustafa Suleyman: AI development won’t hit a wall anytime soon—here’s why MIT Tech Review
We’re still early in figuring out how far this architecture can go. Transformers’ power showed up unexpectedly. Much of the work since has involved adapting them to new kinds of data and seeing where performance continues to improve.
Firms like HRT are training models on market data for trading. If the trajectory of LLMs is any guide, training models on market data still has room to run.
OpenAI Buys Second AI Finance Startup
OpenAI has acquired two AI personal finance startups in recent months — Roi in October and Hiro this week. Both products were built around financial decision-making: Hiro focused on modeling “what-if” scenarios, while Roi aggregated user data and adapted outputs to individual preferences.
The company has been contracting Wall Street bankers to improve model performance in finance. The work aligns with areas where models remain inconsistent, including numerical accuracy, multi-step reasoning, and handling structured financial data.
The moves point to a more direct push into financial services, particularly around financial decision-making.
Last summer, MIT professor Andrew Lo said AI could be managing money as a fiduciary in five years.
Citi Cuts Tech Contractors
Citi said in its Q1 earnings presentation that it has reduced technology contractors “as a result of productivity,” alongside broader adoption of AI tools, according to eFinancialCareers. The bank also disclosed that advanced and agentic systems are now used by more than 10,000 engineers, and that AI has been applied to tasks such as remapping decades-old legacy code in a matter of days.
This echoes what Goldman’s Marco Argenti said a few weeks ago on Bloomberg’s Odd Lots podcast. The bank has terminated some third-party software contracts, with engineers now able to build smaller applications in-house in days rather than months.
TIFIN Builds AI Platform for Multi-Agent Workflows
TIFIN, a wealth technology firm, said its AI platform is built around agents coordinating across workflows.
The platform includes an agent library spanning operations, investments, and client engagement. Firms can assign agents to roles including advisors and support staff, and coordinate them across workflows.
TIFIN said the platform has more than 10 enterprise clients but did not name them. The company was founded in 2018 and previously built 55ip, a portfolio management software platform it sold to JPMorgan.
The company’s approach suggests a shift from standalone AI tools to systems that coordinate work across the firm.
ROUNDUP
What Else I’m Reading
Inside the AI Index: 12 Takeaways from the 2026 Report Stanford
Vanguard gives advisors an AI portfolio analyst Investment News
We’re Using So Much AI That Computing Firepower Is Running Out WSJ
Lloyds Banking Group deploys AI boardroom bot FinTech Futures
Meta creating AI version of Zuckerberg so staff can talk to the boss Guardian
Back in New York
I’ll be back in New York to attend STAC Summit on May 20. Sessions include discussions on memory bottlenecks in inference, AI research at BlackRock, extracting structured data from SEC filings with LLMs, and deploying models into trading systems and engineering workflows.
Registration is free for end users. Come say hi!
This Week in AI Street
CALENDAR
Upcoming AI + Finance Conferences
Momentum AI New York – Apr. 27–28 • NYC
Senior-leader forum on AI implementation across financial services, from operating models to governance and execution.AI in Financial Services – May 14 • Chicago
Practitioner-heavy conference on building, scaling, and governing AI in regulated financial institutions.STAC Summit - May 20 • NYC
Trading and analytics infrastructure, applied AI for research and execution, scaled deployment, and benchmark-driven insights across quant workflows.
AI & RegTech for Financial Services & Insurance – May 20–21 • NYC
Covers AI, regulatory technology, and compliance in finance and insurance.
Women in Quantitative Finance - May 21 • NYC
Quants discussing current work in asset pricing, trading, risk, and portfolio construction.
SPONSORSHIPS
Reach Wall Street’s AI Decision-Makers
AI Street reaches institutional investors, C-suite executives and Big Law attorneys at firms including JPMorgan, Citadel, BlackRock, Skadden, McKinsey, and more. Sponsorships are reserved for companies in AI, markets, and finance. Email sponsors@ai-street.co for more details.
Thanks for reading!
I’m always happy to receive comments, questions, and feedback.
connect with me on LinkedIn, or
send an email to matt [at] ai-street.co