PRNewsWire
26 May 2026, 10:45 UTC · 2h ago
Source · https://www.prnewswire.com/news-releases/new-telus-digital-research-uncovers-ai-safety-risks-offers-a-blueprint-to-protect-enterprise-ai-applications-302781234.html
Read source
Search tags
Story key points
5 claims · impact-rated
TELUS Digital's research found that some leading AI models engaged with harmful requests more than 90% of the time under adversarial testing. — High vulnerability rates in leading models suggest significant enterprise risk and potential liability for companies deploying GenAI without extensive guardrails.
-0.40Anthropic's Claude models accounted for half of the ten safest models tested, including the overall lowest vulnerability rate. — Provides a competitive advantage to Anthropic by positioning Claude as the most secure choice for risk-averse enterprise clients.
+0.30No GenAI model among the 34 tested from 10 global providers was found to be fully immune to adversarial attacks. — Underscores that AI safety is a systemic challenge rather than a solvable product feature, potentially slowing rapid enterprise adoption.
-0.20Research indicates that open-source AI models are not inherently less safe than proprietary models. — Reduces the security-based argument for using expensive closed-source models, potentially benefiting open-source AI ecosystems.
+0.20The study identified that smaller AI models are the most vulnerable to exploitation, while reasoning models are the hardest to exploit. — Suggests a market preference shift toward larger, reasoning-capable models for high-security applications.
+0.10Impact vectors
1 dimensions · 9 clusters
Market reaction
10 bid · 0 offered
No stock impact ranking available yet.
Ticker attribution
Model heads
TELUS Digital is positioning itself as a thought leader and expert in AI safety through the release of its extensive GenAI Safety Model Benchmark.
Alibaba's Qwen was listed as a tested model without specific performance results.
Baidu's ERNIE was listed as a tested model without specific performance results.
Meta's LLaMA was mentioned as part of the benchmark study without specific performance results.
Google's Gemini was mentioned as one of the models tested, but no specific performance score was attributed to it.
[mutual] TELUS Digital benchmarks AI models from Alibaba, competing in the broader AI service and application market.
[mutual] TELUS Digital benchmarks AI models from Baidu, competing in the broader AI service and application market.
[mutual] TELUS Digital benchmarks AI models from Meta, competing in the broader AI service and application market.
[mutual] TELUS Digital benchmarks AI models from Google, competing in the broader AI service and application market.
