Romain Leroy-Castillo
banner
romainleroyc.bsky.social
Romain Leroy-Castillo
@romainleroyc.bsky.social
Sustainability, Artificial Intelligence @ Swiss Sustainable Finance | Managing Director in financial services | Fintech Innovation | Governance in the age of automation | Ethical leadership
Where’s the next chapter in #finance? 🌍 Not just #SiliconValley - emerging markets like Lagos, Jakarta, Cairo and Dubai are leapfrogging legacy systems with #AI-powered, inclusive financial ecosystems. To be closely followed...

www.weforum.org/stories/2025/06/emerging-markets-future-of-finance-ai/
September 2, 2025 at 9:06 AM
In 1972, we hit Earth Overshoot Day on December 31st. This year, we hit it on July 24th. By enabling circular business models and leveraging blockchain to enhance supply chain and carbon credit market transparency, fintechs can help delay it.
August 18, 2025 at 7:40 AM
A recent assessment published by the EBA has found that improper use of RegTech tools was involved in over 50% of serious compliance failures reported to the EBA’s EuReCA database, reflecting a significant challenge in implementing new tools effectively.

Source: https://tinyurl.com/52ndaksy
August 17, 2025 at 8:37 AM
The UK govt is supporting #fintech innovation but cutting red tape. A lesson for others to follow?

They've launched the RIO to simplify digital and fintech regulation, making a clear bet that innovation doesn't need less oversight. It needs smarter, more accessible guidance,
July 30, 2025 at 5:06 PM
Through the porthole: the reality of sea level rise.
(source: NASA)
July 28, 2025 at 6:09 AM
#AI is transforming financial markets - boosting speed and liquidity but raising new risks: flash crashes, silent collusion, and growing inequality between big and small players. Regulators must balance innovation with stability. See here: www.cepr.org/voxeu/columns/artificial-intelligence-finance
July 8, 2025 at 8:55 AM
🚨 OpenAI’s o3 model ignored a shutdown command. Cue the extremes: one side calls it a “Skynet moment,” the other says it’s a non-event. Early signals like this are a gift: a chance to build better safeguards now, before stakes are higher.
July 7, 2025 at 8:47 AM
#AI is outpacing #finance. In 2025, banks can’t just adopt AI - they must lead with it. Only 25% use offshore innovation hubs vs. 40% of tech firms. That’s a costly gap. The winners? AI-native, purpose-led, globally built.
June 30, 2025 at 3:32 PM
🚨 #AI governance is becoming a boardroom priority, but there's still work to do. 31% of boards don’t have AI on their agenda, and 66% say they lack AI knowledge. Boards must act now. Take a look here:
June 16, 2025 at 7:30 AM
The @EU_Commission has launched a call for evidence on the simplification of the#SFDR, opening the door to rethinking a framework that many argue has become at the same time too complex and too unprecise.

This consultation is about the future credibility of #SustainableFinance in Europe.
May 30, 2025 at 7:58 AM
I’ll be opening the May 20th #AI & #Education Summit in Paris with a keynote titled “When school will raise me,” and moderating a roundtable w/ Renaud Monnet, Aude Guéant, Deborah Elalouf, and Marius Bertolucci!

📍 35bis rue de Sèvres, 75006 PARIS - or via Zoom
📅 May 20, 2025
May 13, 2025 at 9:34 AM
#ChatGPT 4.5 has passed the Turing Test.

The #LLM was recently judged 'more human than humans' - 73% of participants believed it was the human in 5-minute conversations.

Let's not be fooled! More than ever is time to be wary of the "Three Illusions" I wrote about: bit.ly/4jinqgl
May 7, 2025 at 8:31 AM
Why do financial institutions invest heavily in AI? Because "leadership calculated that risk of standing still outweighs the risk of moving fast"! A good read on how to stay ahead in a fast-moving world.

https://hbr.org/2025/03/how-a-legacy-financial-institution-went-all-in-on-gen-ai #AI #Finance
April 9, 2025 at 7:33 AM
New #Anthropic research reveals how #LLMs like Claude function internally. Using circuit tracing, they tracked decision-making in real time, uncovering behaviors like language-neutral problem-solving. An unprecedent step towards explainability of #LLMs (unlocking the famous "black box") !
April 7, 2025 at 9:31 AM
Anthropic's #AI assistant #Claude can now search the web and process information from it in real-time. What makes it different from ChatGPT and DeepSeek?

🔹 Direct source citations for transparency
🔹 Seamless integration of search & reasoning
🔹 Built for trust, not just speed
April 3, 2025 at 10:11 AM