Credit & Surety: AI making a difference

AXA XL market survey and webinar explore AI usage

In Q3 2024, AXA XL surveyed Credit and Surety professionals from around the world to gather their perspectives on market trends and drivers since the pandemic, and to find out the extent to which their companies are using or planning to use AI, and for what tasks. The survey and subsequent June 2025 webinar discussion revealed that AI is widely embedded in and benefiting the Credit and Surety sector, that AI will be a disruptor for the sector within the next few years, and that players without AI capabilities will lose out. At the same time, oversight and a secure data environment are vital.

Majority using, or on cusp of using, AI

The survey revealed a spread of usage levels across the sector: 43 % of the survey respondents’ companies are using AI in at least one business area (with 58 % of these companies also planning new projects or the ongoing development of their existing AI tools); 32 % are close to implementing AI in at least one business area or are investigating and/or testing its performance; 25 % are neither using AI nor investigating its potential usage. That 75 % of respondents’ companies are using, or close to using, AI technologies, indicates a high AI value expectation across the sector.

AI usage in the Credit and Surety market. Source: AXA XL Market Survey 2025.

Third-party providers dominate development

Of the survey respondents reporting AI projects, 25% said that their companies are developing solutions purely in-house, while the remainder are working with third-party providers. Interestingly, this decision did not align to company size – it is not just some of the bigger players that have chosen to develop AI systems purely in-house, some smaller players described small teams of existing employees with operational experience and strong technical skills quickly producing effective AI platforms.

Underwriting and risk assessment top the AI-usage chart

The most frequently reported AI usage area was underwriting and risk assessment, followed by operations, risk management and claims management. Overall, survey respondents spoke of improved underwriting decision-making, new data insights, faster analytics and enhanced efficiency. One survey respondent noted that AI has helped to improve their client retention levels by lowering volatility and reducing the need to restrict limits. AI helping to identify new business opportunities was another notable benefit discussed in the survey webinar.

Current AI usage and AI projects in the pipeline, by business area. Percentages shown indicate the percentage of companies reporting AI usage or planned usage for that business area – as some companies reported multiple usage/ planned usage areas, the total exceeds 100 %. Source: AXA XL Market Survey 2025.

Usage examples of AI-supported software:

• Original underwriting risk assessment, rating, risk rate scoring, credit analysis and comparisons
• Fully automated risk assessment of smaller risks/ small bonds
• Creating underwriting memos from complex submissions
• Data transfer from submissions into systems (data scraping)
• Coding support to build out IT infrastructure
• Reducing low-grade administrative tasks
• Data management to upgrade legacy systems
• Monitoring credit ratings to identify patterns and improve portfolio modelling
• Fraud detection, e.g., by spotting unusual movements in financial statements
• Broker tender texts and credit models to assess optimal reinsurance structures for an insurer’s portfolio.

Keeping a close eye on potential threats

Survey respondents highlighted the potential AI threats of data security, accuracy/ hallucinations and data needs/ data access issues for system improvement. It was stressed that experts still need to look critically at the outputs. Panelists in the survey webinar discussion added that where AI is in usage, a secure, controlled data environment is vital, as is a deep understanding of the tool and basis of its outputs.

Where AI is in usage, a secure, controlled data environment, is vital, as is a deep understanding of the tool and basis of its outcomes.

Keeping a close eye on potential threats

Survey respondents highlighted the potential AI threats of data security, accuracy/ hallucinations and data needs/ data access issues for system improvement. It was stressed that experts still need to look critically at the outputs. Panelists in the survey webinar discussion added that where AI is in usage, a secure, controlled data environment is vital, as is a deep understanding of the tool and basis of its outputs.