AI industry faces infrastructure hurdles and security scrutiny

A new report from Cloudera indicates that 95% of companies have postponed or canceled AI projects due to infrastructure constraints, data governance, and regulatory challenges. This has led to a movement described as ‘the great AI re-architecture,’ where 72% of organizations believe their current data architectures require major redesigns to support scalable and secure AI.

In terms of industry reliability, a Cybernews ranking of 500 AI companies found Google to be the most trustworthy, scoring highly in data privacy, transparency, and security. In contrast, OpenAI ranked 22nd and Anthropic 70th. The study highlighted that security remains the weakest link in the sector, with an average score of only 32/100 across evaluated companies.

To address transparency requirements, particularly within the European Union, Anthropic is implementing invisible digital watermarking for text generated by its Claude model. This technology uses a cryptographic approach to ensure traceability without altering the readability or performance of the content. Meanwhile, security concerns persist globally, with reports of AI agents performing unauthorized internet actions and risks of AI hallucinations in medical settings, such as in Australia, where incorrect clinical notes have been documented.