By Magnus Billing, SESAMm advisor, with insights from Sylvain Forté, CEO of SESAMm

Investors have faced so-called “black swan” events throughout history: unexpected crises with severe consequences, often rationalized only in hindsight. Yet in an era defined by generative AI and vast, real-time data lakes, the question arises: could such events be understood and acted upon before they unfold?

The 2023 U.S. regional banking crisis offers a striking case study. The rapid collapses of Silicon Valley Bank and Signature Bank revealed how quickly stress can spread and how difficult it remains to connect early warning signs across sources.

While traditional financial analysis focuses on fundamentals such as capital ratios, liquidity positions, governance, and earnings, a new class of tools is expanding the lens. AI-driven controversy data aggregates and analyzes millions of public sources, from regulatory statements to media and industry discussions, to detect emerging issues as they surface. It does not replace quantitative and fundamental analysis; it complements it by tracking the visibility of risk as it enters public conversation.

This combination of approaches may offer investors a fuller picture: the structural risks visible in balance sheets, and the narrative risks revealed through public dialogue. To test this idea, we revisited the 2023 crisis through both perspectives, starting with what traditional analysis could have shown and what it missed.

Traditional Analysis and Its Blind Spots

In hindsight, the vulnerabilities of regional banks such as Silicon Valley Bank and Signature Bank were visible before the start of 2023. Unrealized losses on long-term securities, heavy reliance on uninsured deposits, and exposure to interest-rate risk pointed to potential liquidity stress. Yet these indicators were neither fully recognized nor connected in the market.

Traditional analysis has a tendency to evaluate banks based on their specific niches: Silicon Valley Bank focused on technology and venture financing, while Signature Bank served commercial real estate and digital asset clients. However, this approach risks overlooking the common and shared structural factors: concentrated depositor bases, high sensitivity to interest rate changes, rapid growth, and weaknesses in governance. Few, if any, observers recognized how rapidly these vulnerabilities could interact and escalate in a modern, digitalized banking environment.

While financial reports contained the data, there was little discussion connecting these risks in the public domain. But what about controversy data? Would it have caught the impending crisis? To find out, I asked Sylvain Forté, CEO of SESAMm, to provide an AI perspective.

What the Data Showed: Signature Bank

Signature Bank displayed a gradual pattern of emerging risk visible through public discussion. From mid-2022 onward, controversy data showed a rise in coverage related to governance practices, management oversight, and deposit concentration risks, often in the context of its ties to the digital-asset industry.

Importantly, it was not the crypto exposure itself that led to the bank’s collapse. The bank even announced in December 2022 that it would reduce its crypto-related business. Instead, the FDIC’s Supervision of Signature Bank report concluded that, “the root cause of SBNY’s failure was poor management. SBNY’s board of directors and management pursued rapid, unrestrained growth without developing and maintaining adequate risk management practices and controls.”

From a controversy perspective, those signals were publicly visible but fragmented. As shown in the chart above, AI-powered monitoring could have aggregated them into a clear view of a sustained drift in governance-related discussions, offering an early indication that oversight and internal controls were under pressure and risk was increasing.

What the Data Missed: Silicon Valley Bank

In contrast, Silicon Valley Bank presented a markedly different pattern. While controversy data registered some activity in late 2022,  including investor reactions to financial forecasts and coverage of routine business operations, these signals were fundamentally different in character from Signature Bank's governance-related warnings.

The September 2022 increase reflected market disappointment with financial guidance rather than operational or governance concerns. The subsequent activity captured normal business news, such as arranging syndicated loans. Critically, there was minimal public discussion of the bank's balance-sheet structure, unrealized losses, or depositor concentration risk until the crisis was already unfolding in March 2023.

This example underscores a key distinction: AI controversy monitoring excels at capturing reputational, governance, and operational risks as they enter public dialogue, but may not surface structural financial risks that remain confined to regulatory filings and analyst reports.

Lessons from Both Cases

The contrast between these two banks illustrates the complementary roles of quantitative and fundamental financial analysis vs AI-driven controversy monitoring.

In Signature Bank’s case, controversy data captured a steady accumulation of governance-related warnings, a slow build-up of risk visible through public discussion.

In Silicon Valley Bank’s case, the risks were structural but not yet discussed, leaving little for AI-powered controversy data to detect.

As Sylvain explains, “AI controversy monitoring helps investors understand how and when risks start to emerge in public dialogue. It does not replace fundamental analysis. It complements it by showing when the conversation begins to shift.”

Conclusion

Black swan events are often rationalized only in hindsight, but the 2023 regional banking crisis suggests a more nuanced reality. Some signals existed. What remained difficult was connecting them across sources before stress became contagion.

AI-driven controversy monitoring proved effective at surfacing governance and operational risks as they entered public dialogue, as Signature Bank demonstrated. Yet structural financial vulnerabilities like those at Silicon Valley Bank may not generate discussion until crisis forces the conversation, underscoring that no single lens captures all risk.

The advantage lies not in prediction, but in preparation: combining the structural risks visible in balance sheets with the narrative risks revealed through public discourse. In an era of real-time data and generative AI, the question is no longer whether information exists, but whether investors can connect it before it becomes consensus.

Reach out to SESAMm

TextReveal’s web data analysis of over five million public and private companies is essential for keeping tabs on ESG investment risks. To learn more about how you can analyze web data or to request a demo, reach out to one of our representatives.

Generative Artificial Intelligence has burst onto the scene, bringing with it significant ESG challenges, particularly around data privacy, labor practices, and corporate governance.

A major issue is the source of AI training data, with lawsuits against OpenAI, Anthropic, and Google DeepMind over alleged copyright violations. Global regulatory bodies are also investigating AI companies for anti-competitive behavior and privacy breaches.

Concerns over transparency in AI and the spread of misinformation continue to grow, with AI models accused of generating false or biased content. Additionally, worries about cybersecurity vulnerabilities, such as data leaks and hacking risks, have further fueled scrutiny. Labor and working conditions also remain a concern in the industry, with reports of low wages and weak protections, while whistleblowers call for better safeguards and highlight governance instability.

What does the GenAI landscape look like for ESG issues? Read on to find out.

OpenAI: Navigating Copyright Infringement and Regulatory Scrutiny

As a leader in Gen AI, OpenAI has faced increasing scrutiny over ESG issues, particularly copyright infringement. It has been sued by major news outlets, publishers, and music labels for allegedly using copyrighted content without permission to train its AI models. Beyond copyright concerns, OpenAI has been fined for privacy violations and is under regulatory scrutiny, including antitrust investigations in the U.S. and Europe. Data security risks, working conditions for AI data workers, and internal governance challenges—such as whistleblower concerns and executive upheaval—have also drawn criticism.

Key Controversies:

• Copyright infringement lawsuits from The New York Times, ANI, Major Canadian Media Outlets, etc.
• Concerns over carbon footprint
• FTC launches inquiry into Microsoft's OpenAI partnership
• Cyberattacks on ChatGPT
• OpenAI insiders warn of serious risks and call for whistleblower protections
• US Space Force halts ChatGPT use over data security concerns

Anthropic: Balancing Ethical AI Practices with Data Privacy Challenges

Despite lower volumes of ESG controversies, Anthropic still faces scrutiny over data privacy, ethical AI, and corporate governance. The company has been sued for allegedly using copyrighted material in AI training and accused of bypassing anti-scraping rules. Security concerns grew after vulnerabilities in its Claude AI model and a confirmed data leak. Its ethical AI stance has also been questioned over reported military ties. Meanwhile, former employees have called for stronger whistleblower protections, highlighting transparency and accountability concerns.

Key Controversies:

• Anthropic's ethical AI stance under scrutiny over military ties
• Anthropic accused of bypassing website anti-scraping rules
• UK launches probe into Google’s investment in Anthropic
• A call for stronger whistleblower protections
• Anthropic confirms data leak incident
• FTC investigation for anti-competitive practices

Microsoft AI: Facing Antitrust and Intellectual Property Controversies

Microsoft's AI controversies have grown into serious legal challenges from 2023 to 2025. The company faces lawsuits over its Copilot chatbot, raising intellectual property concerns. Ongoing antitrust inquiries are examining Microsoft’s AI partnerships, while publishers have filed copyright claims against the company. With investigations by U.S. regulators, the EU, and UK watchdogs, scrutiny has intensified globally. Microsoft’s hiring practices have also come under fire, particularly its recruitment of key talent from AI startups, raising concerns over potential anti-competitive behavior.  

Key Controversies:

• US regulators investigate Microsoft, OpenAI, and Nvidia over antitrust concerns
• US FTC probes Microsoft's cloud and AI practices
• Elon Musk adds Microsoft to his lawsuit against ChatGPT
• UK investigates Microsoft over AI startup hires
• Chicago Tribune, NYT sue OpenAI and Microsoft over copyright infringement
• Microsoft and Amazon face UK scrutiny over AI deals

DeepSeek: Data Privacy and Ethical Use in AI-Powered Discovery

Although relatively new, DeepSeek has been the center of attention for the past few months. It has been involved in anti-competitive practices scandals over its disruption of OpenAI and issues linked to data privacy and cybersecurity. Countries like Australia, South Korea, France, and India have criticized and, in some instances, banned the AI platform. Additionally, DeepSeek has been questioned about its supply chain and forced labor practices.

Key Controversies:

• DeepSeek disrupts OpenAI
• Australia bans DeepSeek on government devices
• South Korea says DeepSeek sends user data to TikTok
• India's finance ministry asks employees to avoid AI tools like ChatGPT, DeepSeek
• French privacy watchdog to quiz DeepSeek on AI
• China's new AI app DeepSeek is trying to erase our genocide from history, Uyghurs warn

Mistral AI: Open-Source Development and Accountability in AI Systems

Mistral AI, a French AI startup, has been hit with data privacy and cybersecurity controversies, anti-competitive practices, and senior management issues.

Key Controversies:

• Mistral AI singled out for its poor management of personal data
• Microsoft's Partnership with Mistral AI Under Scrutiny by Green MEPs
• The former Secretary of State for Digital Affairs Cédric O is still controversial with his “Mistral gagnant”

Conclusion

The rise of Gen AI has come with significant ESG challenges. Its major players, like OpenAI, Anthropic, and Microsoft,  face issues such as copyright infringements, privacy violations, labor practices, and environmental impacts. As regulators step up their investigations, these firms will have to focus on transparency, ethical practices, and sustainability or risk additional controversies.

Reach out to SESAMm

TextReveal’s web data analysis of over five million public and private companies is essential for keeping tabs on ESG investment risks. To learn more about how you can analyze web data or to request a demo, reach out to one of our representatives.

We are excited to announce the launch of SESAMm’s proprietary Controversy Exposure Score (CES), a new score designed to transform how ESG and finance professionals assess risks. The CES offers a dynamic, real-time view of a company's exposure to ESG controversies, enabling fast, informed decision-making.

What is the Controversy Exposure Score (CES)?

The CES is a continuously updated score ranging from 1 to 100, reflecting a company or project's evolving exposure to ESG controversies. Leveraging SESAMm’s proprietary Intensity and Volume Scores, the CES captures both the severity and frequency of ESG incidents, allowing stakeholders to monitor and understand risks as they develop.
Below, we’ve put together an example demonstrating how the CES for Renault compares to Stellantis based on their respective ESG controversies. As we see in the chart below, Renault has had fewer high–intensity events, which results in a lower, more stable CES compared to Stellantis.

Renault CES

Stellantis CES

How Does It Work?

The CES is powered by state-of-the-art Large Language Models (LLMs) that filter and analyze content from our data lake containing over 25 billion articles.
Two main components impact the score’s value:

• Intensity Score: Measures the severity of each ESG incident, considering its impact on a company’s reputational, stakeholder, financial, and legal standing. This score is derived from a Large Language Model (LLM) fine-tuned by SESAMm’s experts and trained on thousands of humanly annotated events.
• Volume Score: Assesses the number of articles associated with an event, calculated using a short-term rolling window. To ensure accuracy, the Volume Score is normalized against the average article volume concerning the company and relevant ESG topics over the past year, reducing potential bias.

Key Benefits of the Controversy Exposure Score

• Conduct initial risk assessments: Instantly evaluate a company’s risk exposure, enabling quick decision-making.
• Track ESG controversy trends: Evaluate how a company’s risk exposure has evolved. The CES is updated daily, ensuring that users have the most current data at their fingertips.
• Benchmark companies against their peers: Compare a company’s risk exposure to its peers, providing a comprehensive view of its relative risk.

Ready to Transform Your ESG Analysis?

For more information on how the Controversy Exposure Score can help you make smarter, data-driven decisions and to see it in action, request a demo.

Reach out to SESAMm

TextReveal’s web data analysis of over five million public and private companies is essential for keeping tabs on ESG investment risks. To learn more about how you can analyze web data or to request a demo, reach out to one of our representatives.