Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
4. Introduction
On April 14, 2026, Goldman Sachs CEO David Solomon issued an unprecedented warning about the cybersecurity risks posed by Anthropic’s Claude Mythos AI model. Speaking on an earnings call, Solomon said the bank is “hyper‑aware” of the enhanced capabilities of frontier AI models and is working “closely” with Anthropic and its security vendors to defend against emerging threats. This Goldman Sachs CEO warning came just days after an emergency meeting convened by US Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell, where bank CEOs were instructed to bolster their cyber defences. This article provides a full analysis of Solomon’s comments, the Treasury‑Fed meeting, and what it means for the financial sector.
For a complete overview of the model itself, read our main guide: Goldman Sachs ‘Hyper‑Aware’ of Anthropic Mythos AI .
5. Main Content
H2: David Solomon’s Earnings Call Comments – Full Transcript Analysis
During Goldman Sachs’ Q1 2026 earnings call, Solomon addressed the Mythos AI risk directly. Key excerpts:
“Obviously the LLMs are making rapid progress and we’re hyper‑aware of the enhanced capabilities of these new models with the help of the US government and the model publishers.”
“We’re working closely with Anthropic and all of our security vendors to kind of harness frontier capabilities wherever it’s possible. And this will continue to be an important focus.”
“We are very focused on supplementing our cyber and infrastructure resilience. And this is part of our ongoing capabilities that we have been investing in, and are accelerating our investment in.”
Solomon did not disclose specific technical details but made clear that Goldman Sachs already has access to Mythos Preview under Project Glasswing and is using it defensively to test its own systems.
For more on Project Glasswing, see our Project Glasswing Deep Dive .
H2: The Treasury‑Fed Emergency Meeting
Just days before Solomon’s comments, Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell summoned the CEOs of systemically important banks – including Solomon – to Washington. The purpose: to discuss the systemic cyber risks posed by advanced AI models like Mythos.
According to sources familiar with the meeting, the message was blunt:
- AI‑powered attacks could occur at machine speed, overwhelming human‑led defences.
- Financial institutions must accelerate investment in automated detection and response.
- Information sharing among banks and with government agencies is critical.
- Mythos, while restricted, is a proof of concept; similar capabilities will become widely available.
The meeting also encouraged banks to join defensive initiatives like Project Glasswing.
H2: Why Goldman Sachs Is ‘Hyper‑Aware’
Solomon’s use of the term “hyper‑aware” is telling. It reflects three specific concerns:
- Emergent capabilities – Mythos was not trained for cybersecurity; its hacking skills emerged as a side effect of general reasoning improvements. Future models could be even more capable.
- Speed of attack – The UK AISI found that Mythos could execute a 32‑step network attack in 2–5 minutes, compared to 20 hours for a human. Banks cannot respond at that pace without automation.
- Autonomous targeting – AISI observed Mythos autonomously targeting “small, weakly defended systems”. While large banks have robust defences, their supply chain and third‑party vendors may not.
For a deeper look at the AISI findings, read our UK AISI Mythos Test deep dive.
H2: How Goldman Sachs Is Responding
Solomon outlined a multi‑pronged response:
- Access to Mythos – Goldman is using Mythos Preview to probe its own systems for vulnerabilities.
- Partnerships – Working with Anthropic, CrowdStrike, Palo Alto Networks, and other security vendors.
- Accelerated investment – Increasing spending on cyber resilience, including AI‑powered defence tools.
- Government coordination – Collaborating with the US Treasury and Federal Reserve on threat intelligence sharing.
Goldman has also reportedly increased its bug bounty payouts and is conducting red‑team exercises using AI‑generated attack scenarios.
H2: The Broader Financial Sector Response
Goldman is not alone. JPMorgan Chase CEO Jamie Dimon also attended the Treasury‑Fed meeting and has since announced a $500 million AI cyber defence initiative. Other major banks, including Bank of America, Citigroup, and Wells Fargo, are following suit.
The Bank of England and Bank of Canada have issued parallel warnings to their respective financial sectors. Global coordination is unprecedented.
H2: Comparison Table – Bank Responses to Mythos AI
| Bank | Key Action | Investment |
|---|---|---|
| Goldman Sachs | Access to Mythos, Project Glasswing participation | Accelerated (undisclosed) |
| JPMorgan Chase | AI cyber defence initiative | $500 million |
| Bank of America | Red‑team AI attack simulations | $200 million |
| Citigroup | Threat intelligence sharing with Treasury | $150 million |
| Wells Fargo | Vendor security audits using AI | $100 million |
H2: Real‑World Applications of the CEO Warning
- For bank executives: The warning is a call to action – invest in AI defences now or risk catastrophic breaches.
- For regulators: The Treasury‑Fed meeting sets a precedent for direct intervention on emerging tech risks.
- For security vendors: Demand for AI‑powered detection and response tools is skyrocketing.
- For the public: While Mythos is restricted, the warning signals that future AI models may be even more dangerous, making personal cyber hygiene essential.
6. External Links (3 real, high‑authority sources embedded above)
- Goldman Sachs Q1 2026 earnings call transcript – David Solomon’s full comments on Mythos AI.
- US Treasury statement on emergency meeting – Official release about the Bessent‑Powell meeting with bank CEOs.
- Financial Times – Banks race to defend against AI cyber attacks – Coverage of industry responses.
7. Internal Links Summary (Outbound from this Supporting Post)
| Anchor Text | Target Slug | Placement |
|---|---|---|
| Goldman Sachs ‘Hyper‑Aware’ of Anthropic Mythos AI | /goldman-sachs-anthropic-mythos-ai-risk | Introduction (inbound to pillar) |
| Project Glasswing Deep Dive | /project-glasswing-cybersecurity | After Solomon’s comments section |
| UK AISI Mythos Test | /uk-aisi-mythos-test | After “Why Goldman Sachs Is ‘Hyper‑Aware’” section |
8. FAQ Section (4 Questions)
Q1: What did Goldman Sachs CEO David Solomon say about Mythos AI?
A: Solomon said the bank is “hyper‑aware” of the cyber risks, working closely with Anthropic, and accelerating investments in cyber resilience.
Q2: Why did the Treasury and Fed call an emergency meeting?
A: To warn bank CEOs about AI‑powered cyber threats, especially after the UK AISI demonstrated Mythos’s ability to autonomously attack networks.
Q3: Is Goldman Sachs using Mythos itself?
A: Yes. Under Project Glasswing, Goldman has restricted access to Mythos Preview and is using it defensively to test its own systems.
Q4: Will other banks follow Goldman’s lead?
A: Yes. JPMorgan Chase, Bank of America, Citigroup, and Wells Fargo have all announced major AI cyber defence initiatives.
9. Conclusion
The Goldman Sachs CEO warning marks a turning point in how the financial industry views AI‑powered cyber threats. David Solomon’s “hyper‑aware” comment reflects a new reality: frontier AI models can attack at machine speed, autonomously find decades‑old bugs, and chain exploits in ways human testers cannot. The Treasury‑Fed emergency meeting underscores that systemic risk is no longer theoretical. Banks are now racing to invest in AI defences, and Project Glasswing provides a model for controlled, defensive use of powerful AI. The question is whether the financial sector can patch its vulnerabilities faster than AI finds them.
Next step: Explore the broader implications of AI for cyber defence and offence in our AI Cyber Defence vs Offence comparison.
