- Today
- Holidays
- Birthdays
- Reminders
- Cities
- Atlanta
- Austin
- Baltimore
- Berwyn
- Beverly Hills
- Birmingham
- Boston
- Brooklyn
- Buffalo
- Charlotte
- Chicago
- Cincinnati
- Cleveland
- Columbus
- Dallas
- Denver
- Detroit
- Fort Worth
- Houston
- Indianapolis
- Knoxville
- Las Vegas
- Los Angeles
- Louisville
- Madison
- Memphis
- Miami
- Milwaukee
- Minneapolis
- Nashville
- New Orleans
- New York
- Omaha
- Orlando
- Philadelphia
- Phoenix
- Pittsburgh
- Portland
- Raleigh
- Richmond
- Rutherford
- Sacramento
- Salt Lake City
- San Antonio
- San Diego
- San Francisco
- San Jose
- Seattle
- Tampa
- Tucson
- Washington
Developers Grapple with AI Trust Gap as Security, Memory, and Interoperability Concerns Persist
Findings from the annual Stack Overflow developer survey reveal growing pains as AI becomes ingrained in enterprise operations.
Published on Feb. 22, 2026
Got story updates? Submit your updates here. ›
The annual developer survey commissioned by Stack Overflow found that while 84% of developers now use AI, only 29% trust the accuracy of AI tools. This highlights the growing challenges around issues such as security, memory, cost, and interoperability that are causing developers to question the presumption that AI is a revolutionary productivity tool. Developers are grappling with problems like the memory-bound nature of AI, the need for smaller models, and the lack of governance controls for Model Context Protocol (MCP) servers. Interoperability also emerges as a key concern, with IBM and others emphasizing the importance of a control layer for agent-to-agent interaction.
Why it matters
As AI becomes more deeply embedded in enterprise operations, the lack of trust from developers poses a significant obstacle to widespread adoption. Resolving issues around security, memory, cost, and interoperability will be crucial for AI to fulfill its promise as a transformative technology. The developer community's concerns reflect the broader challenges the industry faces in building reliable and trustworthy AI systems.
The details
Developers are facing several key challenges with AI, including the memory-bound nature of AI systems, which require about eight times more memory than traditional machines to deliver accurate results. To address this, companies like Oracle are providing representations of memory as tables in a database, allowing developers to build agents that can remember. Another solution is the move towards smaller models, which can reduce memory load and cost. However, many developers are still focused on accessing leading large language models from companies like OpenAI and Anthropic. Another area of concern is the security and governance of Model Context Protocol (MCP) servers, which provide LLMs and AI agents with the ability to connect to external data sources and applications. Security professionals have documented issues with MCP servers lacking proper authentication and access controls, leading companies like Descope and WSO2 to develop solutions to address these governance challenges. Interoperability also emerges as a key issue, with IBM emphasizing the need for a control layer to manage agent-to-agent interactions.
- The annual developer survey commissioned by Stack Overflow was conducted in 2026.
The players
Stack Overflow
A popular web resource in the developer community that commissioned the annual developer survey.
Tony Loehr
A solutions engineer at Cline Bot Inc. who spoke at the Developer Week conference about the challenges of using AI.
Richmond Alake
The director of AI developer experience at Oracle Corp. who spoke at the Developer Week conference about the importance of memory for AI developers.
Legare Kerrison
A developer advocate for AI at Red Hat who spoke at the Developer Week conference about the benefits of smaller AI models.
Jody Bailey
The chief product and technology officer at Stack Overflow who was interviewed by SiliconANGLE during the Developer Week conference.
What they’re saying
“If AI is supposed to be a revolutionary productivity tool, then why am I still doing most of the work?”
— Tony Loehr, Solutions Engineer, Cline Bot Inc. (SiliconANGLE)
“The nature of what we have to do to build AI agents in production is changing. We need memory to be front and center.”
— Richmond Alake, Director of AI Developer Experience, Oracle Corp. (SiliconANGLE)
“When the LLMs are smaller, it means that our costs are also smaller and we are moving faster. There's less time to first token.”
— Legare Kerrison, Developer Advocate for AI, Red Hat (SiliconANGLE)
“Right now, it's hard to bet against the foundation models, especially in our space. I do believe there are lots of places where small models make a ton of sense.”
— Jody Bailey, Chief Product and Technology Officer, Stack Overflow (SiliconANGLE)
“You can't rely on agents to govern themselves. Your gateway might need to evolve.”
— Derric Gilling, Vice President and General Manager of the API Platform, WSO2 (SiliconANGLE)
What’s next
The developer community is closely examining the merits of both small and large AI models, and this situation is likely to become clearer over the coming year. Additionally, companies are working to address the security and governance challenges around Model Context Protocol (MCP) servers, with solutions like those from Descope and WSO2 aiming to provide more secure and controlled access to these servers.
The takeaway
The growing pains experienced by developers as AI becomes more deeply integrated into enterprise operations highlight the critical need to address issues around security, memory, cost, and interoperability. Resolving these challenges will be essential for AI to fulfill its promise as a transformative technology and gain widespread trust and adoption among developers and the broader business community.
San Jose top stories
San Jose events
Mar. 11, 2026
San Jose Barracuda vs. San Diego GullsMar. 12, 2026
Phil HanleyMar. 13, 2026
Phil Hanley




