📢 AI in February 2025: The Breakthroughs, Regulations, and What Comes Next

⏳ Estimated Reading Time: 11 minutes

Artificial Intelligence is evolving at an unprecedented pace. In the past month alone, we’ve seen:

🔹 A major global market disruption from DeepSeek
🔹 New reports reshaping AI governance and cybersecurity
🔹 And another part of European AI Act is entering into force on February 2nd

Let’s break it down and look at what needs to happen next.

1. DeepSeek: The AI Shaking Markets & Security

The arrival of DeepSeek, a Chinese AI model challenging OpenAI and Google, marked one of the most dramatic market shifts in AI history.

Key Takeaways:

📉 $593 billion wiped from AI-related stock value, as Nvidia, Microsoft, and Google tumbled.
📈 DeepSeek R1 is 20–50x cheaper to run than OpenAI’s GPT-4.
📲 DeepSeek became the #1 AI app in the US, ahead of ChatGPT.
🛑 DeepSeek is already banned in Italy over data security concerns .

Security implications emerged quickly:

🚨 Massive cyberattacks (or so called by the company) hit DeepSeek’s servers.
🔓 A critical data leak exposed API keys, infrastructure details and conversations of users with the system.
🇨🇳 China’s AI dominance is now a geopolitical priority, which may become a reality in the next 2 to 5 years.

What’s next? Tighter US export bans, AI chip production shifts, and a reevaluation of AI governance worldwide.

2. AI Safety & Workforce Trends: Reports You Need to Know

A wave of new AI research hit this month, changing how we think about security, jobs, and governance.

📌 International AI Safety Report (January 2025)

🔹 Led by Prof. Yoshua Bengio, this report highlights cyber risks, AI-generated misinformation, and labor market shifts .

🔹 AI “agents” that act with minimal human oversight pose emerging security challenges.

🔹 Calls for international AI risk management standards.

The Report

📌 LinkedIn Work Change Report

🔹 By 2030, 70% of workplace skills will change, driven by AI .

🔹 AI Engineer is one of the fastest-growing roles in 15 countries.

🔹 88% of executives say AI adoption is their top priority this year.

LinkedIn’s report

📌 World Economic Forum AI Security Report

🔹 Cyber threats are rising with AI adoption .

🔹 AI systems are targets for sophisticated attacks, from data poisoning to adversarial manipulation.

🔹 Urges businesses to adopt “Shift Left” security strategies (building security in from development).

The WEF report on AI and Cybersecurity

📌 France’s New AI Institute (INESIA)

🇫🇷 France just announced a National AI Security & Evaluation Institute. Goal: Strengthen AI security, testing, and regulation. It will be hosted at the SGDSN.

🔹 Collaborating with ANSSI (national cybersecurity agency), Inria, and European AI offices.

The press release

3. A major milestone of the EU AI Act Is Now in Force (Feb 2nd, 2025)

📜 The first major AI law in the world just went live. The EU AI Act creates a risk-based framework :

🟥 Banned AI: Social scoring, predictive policing, and manipulative AI (Article 5).

🟧 High-Risk AI: Healthcare, finance, employment—must meet strict security standards.

🟩 Limited-Risk AI: Chatbots & AI tools need transparency rules.

Banned AI becomes into force.

Impact:

🔹Fines up to €35 million (or 7% of global revenue) for violations of Article 5.

🔹AI developers must prove safety before launching new models.

🔹Non-compliance could force AI firms out of Europe.

The Takeaway: AI regulation is no longer theoretical— compliance is now a business requirement. It will be cheaper to adapt now rather than as an afterthought.

4. What Needs to Happen Next?

With AI evolving faster than regulations, businesses and governments must step up security measures.

🔹 Apply WEF AI Safety Report Recommendations

1️⃣ Create “red teams” for AI risk assessment.

2️⃣ Develop early warning systems for AI security breaches.

3️⃣ Use “explainable AI” models to increase transparency .

🔹 Leverage the NIST AI Risk Framework

The NIST AI RMF 1.0 provides a global blueprint for AI risk management, focusing on:

✅ Security-first AI development

✅ Bias mitigation & transparency

✅ Continuous AI risk assessment

It goes deeper than only AI security.

Get the NIST AI RMF

🔹 Strengthen AI Workforce Readiness

📈 Invest in AI upskilling & security training (LinkedIn predicts 70% of jobs will change!).

🔐 Implement “Shift Left” cybersecurity—security must be built into AI, not added later.

5. Call to Action: Your Move in the AI Security Race

The AI arms race has begun. Geopolitics, cybersecurity, and business strategy are colliding. To stay ahead, organizations must act now.

🔹 If you’re a risk, compliance, or data leader
→ Establish an AI Trust Policy.

✅ Define governance and accountability for AI risks.

✅ Ensure compliance with the EU AI Act & NIST AI RMF.

✅ Set transparency, bias, and security standards for AI use.

🔹 If you’re in AI development / data science
→ Align with AI governance frameworks.

✅ Comply with EU AI Act, NIST AI Risk Framework, and emerging AI security standards.

✅ Build secure, explainable, and bias-mitigated AI models.

✅ Ensure data integrity and protect against AI adversarial attacks.

🔹 If you’re in cybersecurity
→ Strengthen AI-specific security controls.

✅ Implement robust AI threat detection & mitigation strategies.

✅ Ensure leadership accountability for AI risks & security compliance.

✅ Leverage AI-powered cybersecurity to defend against increasingly AI-driven attacks.

🔹 If you’re a business leader
→ Define AI responsibility & ensure workforce readiness.

✅ Establish clear AI governance & risk ownership within your organization.

✅ Train teams on AI literacy, compliance, and security risks.

✅ Prepare for workforce transformation—70% of job skills will change by 2030.

👇 What do you think? Drop a comment or forward this to a colleague who needs to see it.

This newsletter was powered by AI and curated, driven and verified by Sylvan Ravinet.

🚀 Share the newsletter👇