# Marco Argenti on AI: Goldman Sachs CIO

Source: https://www.bankingnewsai.com/executives/marco-argenti
Last updated: Nov 21, 2026

Marco Argenti · Chief Information Officer, [Goldman Sachs](https://www.bankingnewsai.com/banks/goldman-sachs) · role as of Nov 2, 2026 [1][13][11]

Marco Argenti is chief information officer of Goldman Sachs, which he joined as a partner in 2019 from Amazon Web Services, and sits on its Firmwide Artificial Intelligence Risk Committee. He rolled out the GS AI Assistant, began piloting Cognition's Devin coding agent in 2025 and, in 2026, is developing Claude-based agents with embedded Anthropic engineers.

## Key points

- Member of the Management Committee and of the Firmwide Artificial Intelligence Risk Committee, per the bank's biography. [1]
- January 2025: the GS AI Assistant reached about 10,000 employees, with all knowledge workers the goal for that year, he told CNBC. [3]
- July 2025: Goldman began piloting Cognition's Devin; he expected hundreds of instances, possibly thousands, beside 12,000 human developers. [6]
- February 2026: he told CNBC that embedded Anthropic engineers had spent six months co-developing agents for trade accounting and client onboarding. [8]

## At a glance

| Fact | Value |
| --- | --- |
| Title | Chief Information Officer [1][13][11] |
| Bank | [Goldman Sachs](https://www.bankingnewsai.com/banks/goldman-sachs) |
| Joined Goldman Sachs | 2019, as a partner [1] |
| Committees | Management Committee; Firmwide Technology Risk, Enterprise Risk and Artificial Intelligence Risk committees [1] |
| Engineers overseen | 12,000 (Fortune, May 2026) [9] |
| AI access | Half of 46,000 employees had AI access (Fortune, March 2025) [4] |

## In Argenti's words

What has Argenti said about AI?

> “You have to embrace the fact that any agent could misbehave.” [11]
>
> — Oct 22, 2026, [CIO Dive](https://www.ciodive.com/news/jpmorgan-chase-goldman-sachs-AI-security/833764/). Evident AI Symposium, New York, reported by CIO Dive

> “I don't think a priori we should block access to any model” [10]
>
> — Sep 8, 2026, [Axios, read on Yahoo Finance](https://finance.yahoo.com/technology/ai/articles/goldman-ai-chief-dont-rule-095705759.html). On open-weight models, told Axios (read on Yahoo Finance)

> “Our philosophy right now is that we're injecting capacity, which in most cases will allow us to do things faster, which translates to a better client experience and more business.” [8]
>
> — Feb 6, 2026, [CNBC](https://www.cnbc.com/2026/02/06/anthropic-goldman-sachs-ai-model-accounting.html). Interview with CNBC on agents built with Anthropic

> “In my 40 years in technology, 2025 saw the biggest changes I have seen in my career” [7]
>
> — Jan 22, 2026, [Goldman Sachs](https://www.goldmansachs.com/insights/articles/what-to-expect-from-ai-in-2026-personal-agents-mega-alliances). Goldman Sachs article, "What to Expect From AI in 2026"

> “Initially, we will have hundreds of Devins [and] that might go into the thousands, depending on the use cases” [6]
>
> — Jul 11, 2025, [CNBC](https://www.cnbc.com/2025/07/11/goldman-sachs-autonomous-coder-pilot-marks-major-ai-milestone.html). Interview with CNBC on the Devin pilot

> “I see evidence that they could be superhumanly productive” [2]
>
> — Jun 27, 2024, [Fortune](https://fortune.com/2024/06/28/goldman-sachs-cio-ai-superhuman-productivity-superhumanly-smart). Interview with Fortune at the Cerebral Valley AI Summit, New York

## AI record

What has Argenti done on AI, and when?

- **Oct 22, 2026** — Contain agents; learn from lab engineers [11][12]. At the Evident AI Symposium he called agents stochastic machines and urged containment, rules for agents and gates on user tools before an agent completes a transaction. He described forward-deployed engineers from AI labs as the fastest way for teams to learn.
- **Sep 8, 2026** — Four-layer safeguards for open-weight models [10]. He told Axios that model testing and certification, secure inference, monitored agent permissions and controlled data access could let banks run any model; Goldman's open-model use is mostly or exclusively U.S.-based.
- **May 8, 2026** — Measures team speed, not individual AI usage [9]. Fortune, citing his Business Insider interview, reported that he judges AI by how fast his 12,000 engineers move from idea to production.
- **Feb 6, 2026** — Claude agents for accounting and onboarding [8]. He told CNBC the bank was in the early stages of Claude-based agents for trade accounting and client vetting and onboarding, and that Devin was broadly available to engineers.
- **Jul 11, 2025** — Devin pilot announced [6]. He said Devin, supervised by staff, would take on work such as updating code to newer languages, and could raise productivity by up to three or four times the rate of earlier AI tools.
- **Mar 19, 2025** — Engineers first, then firmwide access [4]. Fortune reported half of 46,000 employees had AI access; engineers came first, with GitHub Copilot and Gemini Code Assist.
- **Jan 21, 2025** — GS AI Assistant reaches about 10,000 staff [3]. He told CNBC the assistant draws on OpenAI, Google Gemini and Meta Llama models, with Anthropic, Mistral and Cohere under review.
- **Jun 27, 2024** — Copilot in the software life cycle [2]. He told Fortune GitHub Copilot was integrated into the bank's development process, with generated code passing the same quality checks, and would not lead to job cuts.

## Where Argenti stands

| Topic | Position | As of | Sources |
| --- | --- | --- | --- |
| Jobs and workforce | Called it premature to expect AI job losses in compliance and accounting; the aim is added capacity. Argues junior AI-native staff become more critical, as managers of agents. | Feb 6, 2026 | [8][5] |
| Spending and returns | Measured coding-assistant returns by use and code-acceptance rates in 2025; by 2026 judged AI by how fast teams ship rather than by individual usage. | May 8, 2026 | [4][9] |
| Risk and regulation | Treats any model as potentially compromised: zero trust and defence in depth, with AI systems reaching data only through the bank's data platform, and gates before agents transact. | Oct 22, 2026 | [10][11] |
| Build or buy and vendors | Wants to plug and play models from several vendors rather than rely on one, and calls open and closed models equally important to U.S. AI competitiveness. | Sep 8, 2026 | [4][10] |
| AI agents | Sees agents as digital co-workers supervised by people; until they will not act irrationally, humans must stay in the loop. | Feb 6, 2026 | [5][8] |

## Career

- 2019–present: Goldman Sachs, Partner (joined 2019); currently Chief Information Officer [1]
- 2013–2019: Amazon Web Services, Vice President of Technology [1]
- 2011–2013: Nokia Corporation, Senior Vice President and Global Head of Developer Experience and Marketplace [1]
- Earlier: Dada S.p.A., Board member and Chief Executive Officer [1]
- Earlier: Microforum Inc., Board member, EVP of Strategy Development and Chief Technology Officer [1]
- Earlier: Dreamware S.r.l. (sold to Microforum), Founder [1]

Education: Master's degree in Informatics Engineering, Università di Pisa; Professional Engineering qualification in Management Engineering, Politecnico di Milano. [1]

## Common questions

### Who is Marco Argenti?

Marco Argenti is chief information officer of Goldman Sachs and a member of its Management Committee. Argenti joined the firm as a partner in 2019 after six years as vice president of technology at Amazon Web Services.

### What has Marco Argenti said about AI and jobs?

In February 2026 Argenti told CNBC it was premature to expect AI agents to cause job losses in compliance and accounting, describing the aim as injecting capacity. In 2024 Argenti said GitHub Copilot would not lead to job cuts.

### What is Marco Argenti's role in Goldman Sachs' AI strategy?

Argenti sits on the Firmwide Artificial Intelligence Risk Committee and announced the GS AI Assistant rollout, the Devin pilot and the Claude-based agents being developed with Anthropic engineers.

### Does Marco Argenti support open-weight AI models?

Yes, under safeguards. In September 2026 Argenti told Axios no model should be blocked a priori, proposing four layers: model certification, secure inference, monitored agent permissions and controlled data access.

## Sources

1. [Marco Argenti (Management Committee biography; undated page, retrieved 2026-11-21)](https://www.goldmansachs.com/our-firm/our-people-and-leadership/leadership/management-committee/marco-argenti) — Goldman Sachs, Nov 21, 2026
2. [Goldman Sachs CIO says AI tools may offer superhuman productivity, but ‘I don’t know if they can be superhumanly smart’](https://fortune.com/2024/06/28/goldman-sachs-cio-ai-superhuman-productivity-superhumanly-smart) — Fortune, Jun 28, 2024
3. [Goldman Sachs rolls out an AI assistant for its employees as artificial intelligence sweeps Wall Street](https://www.cnbc.com/2025/01/21/goldman-sachs-launches-ai-assistant.html) — CNBC, Jan 21, 2025
4. [Why Goldman Sachs’ CIO is taking a measured approach to rolling out AI across the business](https://fortune.com/2025/03/19/goldman-sachs-cio-ai) — Fortune, Mar 19, 2025
5. [Marco Argenti: We Must Prepare AI Natives to Shape the Future of Work—as Only They Can](https://www.goldmansachs.com/insights/articles/fortune-we-must-prepare-ai-natives-to-shape-the-future-of-work) — Goldman Sachs, Jul 9, 2025
6. [Goldman Sachs is piloting its first autonomous coder in major AI milestone for Wall Street](https://www.cnbc.com/2025/07/11/goldman-sachs-autonomous-coder-pilot-marks-major-ai-milestone.html) — CNBC, Jul 11, 2025
7. [What to Expect From AI in 2026: Personal Agents, Mega Alliances, and the Gigawatt Ceiling](https://www.goldmansachs.com/insights/articles/what-to-expect-from-ai-in-2026-personal-agents-mega-alliances) — Goldman Sachs, Jan 22, 2026
8. [Goldman Sachs taps Anthropic’s Claude to automate accounting, compliance roles](https://www.cnbc.com/2026/02/06/anthropic-goldman-sachs-ai-model-accounting.html) — CNBC, Feb 6, 2026
9. [Goldman Sachs’ tech boss says tracking individual AI usage isn’t useful. He just watches how fast his 12,000 engineers move from idea to production](https://fortune.com/2026/05/08/goldman-sachs-cio-marco-argenti-tech-ai-future-of-work-employees/) — Fortune, May 8, 2026
10. [Goldman AI chief: Don't rule out open models](https://finance.yahoo.com/technology/ai/articles/goldman-ai-chief-dont-rule-095705759.html) — Axios, read on Yahoo Finance, Sep 8, 2026
11. [JPMorgan Chase, Goldman Sachs underscore security in AI plans](https://www.ciodive.com/news/jpmorgan-chase-goldman-sachs-AI-security/833764/) — CIO Dive, Oct 22, 2026
12. [JPMorgan Chase, Goldman Sachs underscore security in AI plans](https://www.bankingdive.com/news/jpmorgan-chase-goldman-sachs-AI-security-lori-beer-marco-argenti/833792/) — Banking Dive, Oct 23, 2026
13. [Goldman restricts traders and bankers from using some AI tools - Bloomberg](https://finance.yahoo.com/news/goldman-restricts-traders-bankers-using-063339234.html) — Bloomberg, read on Yahoo Finance, Nov 2, 2026

## More

- [Goldman Sachs AI strategy](https://www.bankingnewsai.com/banks/goldman-sachs)
- Also at Goldman: [David Solomon, Chairman and CEO](https://www.bankingnewsai.com/executives/david-solomon)
- [All technology chiefs](https://www.bankingnewsai.com/executives#technology)
- [All executive dossiers](https://www.bankingnewsai.com/executives)


---

Canonical page: https://www.bankingnewsai.com/executives/marco-argenti
Part of [BankingNewsAI](https://www.bankingnewsai.com/) — a free daily brief on AI in banking, an AI regulation tracker (60 authorities, 360 documents) and AI-strategy profiles of the 120 largest US banks. Markdown versions of every reference page: append `.md` to the page URL; index at https://www.bankingnewsai.com/llms.txt.
