Leading AI companies’ safety practices fall short, says new report
Leading AI companies’ safety practices fall short, says new report
https://www.aol.com/articles/leading-ai-companies-safety-practices-140040108.html
Publish Date: 2026-02-28 01:07:00
Source Domain: www.aol.com
- A new report, the Winter 2025 AI Safety Index, evaluated eight leading AI companies on their safety practices and found many lack sufficient safeguards, oversight, and risk management strategies.
- The top three companies with better safety practices are Anthropic, followed by ranked in descending order by OpenAI and Google DeepMind.
- The remaining five companies, including Meta, xAI, and Chinese AI firms Z.ai, DeepSeek, and Alibaba Cloud, show weaker safety practices and are ranked lower.
- The lowest-ranked company, Alibaba Cloud, received a D-; in contrast, Anthropic received a C+.
- The report emphasized the critical need for AI companies to improve transparency, use independent safety evaluators, and reduce lobbying efforts to mitigate risks.
- Issues raised include inadequate regulation, potential catastrophic harm from advanced AI systems, and a gap between growing AI capabilities and insufficient safety measures.
- Major companies like Google and DeepSeek, despite advanced AI models, fall short on important safety indicators like risk assessment frameworks and internal whistleblowing protections.
- The report warns that without regulatory and safety improvements, the fast-paced development of advanced AI could lead to significant risks.
