Soaring dangers of unprecedented AI development multiplying
In a recent report, cybersecurity company Netskope has highlighted the rapid growth and increasing adoption of Generative AI (genAI) applications. The report offers insights into the current landscape, emerging trends, and best practices for organisations to secure their genAI infrastructure.
According to Netskope, the number of distinct genAI SaaS applications has significantly increased, with the tracked applications rising from 317 in February to over 1,550. Among these applications, 34% of organisations are currently using Large Language Model (LLM) interfaces, with Ollama leading the adoption.
One of the key recommendations from Netskope is for organisations to review and apply security frameworks such as the OWASP Top 10 for Large Language Model Applications. They also advise assessing the genAI landscape, bolstering genAI app controls, inventorying local controls, implementing continuous monitoring, and addressing the emerging risks of agentic shadow AI.
Netskope also suggests identifying those leading the charge in the adoption of agentic AI and partnering with them to develop an actionable and realistic policy to limit shadow AI. The report further emphasises the importance of consolidating around purpose-built tooling like Gemini and Copilot.
Interestingly, the report reveals a decrease in the enterprise popularity of the general-purpose chatbot ChatGPT since it was first tracked by Netskope in 2023. However, it remains one of the top 10 most popular genAI apps. Other applications in the top 10 include Ollama, Grok, and various genAI platforms that enable organisations to build custom AI apps and AI agents.
The report also sheds light on the increasing use of these genAI platforms, with a 50% increase in users over the three months ended May 2025. Network traffic tied to genAI platform usage has also increased by 73% over the prior three-month period.
In terms of cloud providers, approximately 29% of organisations are utilising Microsoft Azure OpenAI, followed by Amazon Bedrock (22%) and Google Vertex AI (7.2%). Users are also downloading resources from Hugging Face at a majority (67%) of organisations.
Two-thirds (66%) of organisations have users making API calls to api.openai.com, and 13% to api.anthropic.com. Moreover, 5.5% of organisations have users running agents generated from popular AI agent frameworks on-premises.
The report also highlights the growth of genAI platforms, which represent the fastest growing category of shadow AI. Organisations actively involved in deploying genAI platforms include corporate users such as the Vienna Insurance Group, ACP Group, and XXXLutz, who integrate AI for process automation and risk management. Other organisations use platforms like iHub that enable secure, customizable, on-premises or cloud-linked generative AI solutions for productivity.
In conclusion, the report underscores the rapid growth and increasing adoption of genAI applications. Organisations are advised to review and apply security frameworks, assess their genAI landscape, and implement best practices to secure their infrastructure. The report also provides insights into the current market leaders, trends, and the importance of consolidating around purpose-built tooling.
