Autonomous AI adoption stalls amid trust and governance crisis
(2025/10/01)
- Reference: 1759317949
- News link: https://www.theregister.co.uk/2025/10/01/gartner_ai_agents/
- Source link:
Enterprises aren't keen on letting autonomous agents take the wheel amid fears over trust and security as research once again shows that AI hype is crashing against the rocks of reality.
Tech analyst Gartner [1]looked at the next big thing in AI evolution – fully autonomous AI agents – and found that most IT application leaders are still avoiding the technology.
The industry-wide survey of 360 bigwigs from organizations with at least 250 full-time employees found that just 15 percent were considering, piloting, or deploying fully autonomous agents.
[2]
Around three-quarters of respondents were piloting or deploying some form of AI agent in their organization, according to the Gartner's research. However, the figures indicate that few are willing to consider the next step – going fully autonomous.
[3]
[4]
Only 19 percent had high or complete trust in their vendor's abilities to protect against AI hallucinations, and a whopping 74 percent worried that AI agents represented a new attack vector in their organization.
This indicates a widening gap between the hype of autonomous AI and the reality enterprises face when trying to implement it.
[5]
Max Goss, senior director analyst at the consultancy, said: "concerns around governance, maturity and agent sprawl continue to hamper the deployment of truly agentic AI."
A recent [6]study found that more than 40 percent of agentic AI projects would be cancelled by the end of 2027, with rising costs, unclear business value, and insufficient risk controls cited as factors.
Former AI cheerleaders including [7]Klarna and [8]Duolingo reportedly switched back to humans after an attempt to [9]swap meatbags for silicon resulted in a drop in quality.
[10]
Even an AI evangelist such as Salesforce, which [11]slashed thousands of customer support roles in the name of the technology, has produced figures showing that LLM agents [12]aren't great at customer confidentiality or multi-step tasks.
[13]US banking giant Citi pilots agentic AI with 5,000 staff
[14]AI coding hype overblown, Bain shrugs
[15]AI in your toaster: Analyst predicts $1.5T global spend in 2025
[16]Your call is very important to us – which is why we're connecting you to a human
Other large companies are also looking to AI tech to trim staff. UK telecoms giant BT, for example, [17]reckons that it can go [18]beyond the 55,000 employees it wants to cut by 2030 thanks to "the full potential" of AI. Earlier this week, the German Lufthansa Group said it will [19]eliminate 4,000 administrative jobs through "digitalization, automation, and process consolidation."
However, reality and expectations continue to differ. A [20]report in the Financial Times showed there were few signs of AI taking roles from humans, and [21]McKinsey found that flooding an organization with AI agents rarely works; success depends on rethinking workflows and integrating the tech properly.
Gartner notes today that the majority of leaders in its survey didn't expect AI agents to replace applications or workers in the next two to four years.
"Only 12 percent strongly agreed AI agents would replace applications, and just 7 percent strongly agreed they would replace workers in that time frame," it said. ®
Get our [22]Tech Resources
[1] https://www.gartner.com/en/newsroom/press-releases/2025-09-30-gartner-survey-finds-just-15-percent-of-it-application-leaders-are-considering-piloting-or-deploying-fully-autonomous-ai-agents
[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aN1QEwvpKGU-r-lMPx7Q8QAAA1U&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aN1QEwvpKGU-r-lMPx7Q8QAAA1U&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aN1QEwvpKGU-r-lMPx7Q8QAAA1U&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aN1QEwvpKGU-r-lMPx7Q8QAAA1U&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[6] https://www.theregister.com/2025/06/29/ai_agents_fail_a_lot/
[7] https://fortune.com/2025/05/09/klarna-ai-humans-return-on-investment/
[8] https://www.reddit.com/r/Futurology/comments/1kuilps/duolingo_ceo_walks_back_aifirst_comments_i_do_not/
[9] https://www.theregister.com/2024/08/29/ai_hiring_freeze_klarna/
[10] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aN1QEwvpKGU-r-lMPx7Q8QAAA1U&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[11] https://www.theregister.com/2025/09/02/salesforce_4000_jobs_ai/
[12] https://www.theregister.com/2025/06/16/salesforce_llm_agents_benchmark/
[13] https://www.theregister.com/2025/09/24/citi_pilots_agentic_ai/
[14] https://www.theregister.com/2025/09/23/developers_genai_little_productivity_gains/
[15] https://www.theregister.com/2025/09/17/gartner_ai_spending/
[16] https://www.theregister.com/2025/09/12/gartner_support_ai/
[17] https://www.theregister.com/2025/06/16/bt_chief_says_ai_could_cut_more_staff/
[18] https://www.theregister.com/2023/05/18/bt_redundancies_2030/
[19] https://newsroom.lufthansagroup.com/en/lufthansa-group-is-consistently-pursuing-its-strategy-and-aims-to-significantly-increase-profitability/
[20] https://www.ft.com/content/c9f905a0-cbfc-4a0a-ac4f-0d68d0fc64aa
[21] https://www.mckinsey.com/capabilities/quantumblack/our-insights/one-year-of-agentic-ai-six-lessons-from-the-people-doing-the-work
[22] https://whitepapers.theregister.com/
Tech analyst Gartner [1]looked at the next big thing in AI evolution – fully autonomous AI agents – and found that most IT application leaders are still avoiding the technology.
The industry-wide survey of 360 bigwigs from organizations with at least 250 full-time employees found that just 15 percent were considering, piloting, or deploying fully autonomous agents.
[2]
Around three-quarters of respondents were piloting or deploying some form of AI agent in their organization, according to the Gartner's research. However, the figures indicate that few are willing to consider the next step – going fully autonomous.
[3]
[4]
Only 19 percent had high or complete trust in their vendor's abilities to protect against AI hallucinations, and a whopping 74 percent worried that AI agents represented a new attack vector in their organization.
This indicates a widening gap between the hype of autonomous AI and the reality enterprises face when trying to implement it.
[5]
Max Goss, senior director analyst at the consultancy, said: "concerns around governance, maturity and agent sprawl continue to hamper the deployment of truly agentic AI."
A recent [6]study found that more than 40 percent of agentic AI projects would be cancelled by the end of 2027, with rising costs, unclear business value, and insufficient risk controls cited as factors.
Former AI cheerleaders including [7]Klarna and [8]Duolingo reportedly switched back to humans after an attempt to [9]swap meatbags for silicon resulted in a drop in quality.
[10]
Even an AI evangelist such as Salesforce, which [11]slashed thousands of customer support roles in the name of the technology, has produced figures showing that LLM agents [12]aren't great at customer confidentiality or multi-step tasks.
[13]US banking giant Citi pilots agentic AI with 5,000 staff
[14]AI coding hype overblown, Bain shrugs
[15]AI in your toaster: Analyst predicts $1.5T global spend in 2025
[16]Your call is very important to us – which is why we're connecting you to a human
Other large companies are also looking to AI tech to trim staff. UK telecoms giant BT, for example, [17]reckons that it can go [18]beyond the 55,000 employees it wants to cut by 2030 thanks to "the full potential" of AI. Earlier this week, the German Lufthansa Group said it will [19]eliminate 4,000 administrative jobs through "digitalization, automation, and process consolidation."
However, reality and expectations continue to differ. A [20]report in the Financial Times showed there were few signs of AI taking roles from humans, and [21]McKinsey found that flooding an organization with AI agents rarely works; success depends on rethinking workflows and integrating the tech properly.
Gartner notes today that the majority of leaders in its survey didn't expect AI agents to replace applications or workers in the next two to four years.
"Only 12 percent strongly agreed AI agents would replace applications, and just 7 percent strongly agreed they would replace workers in that time frame," it said. ®
Get our [22]Tech Resources
[1] https://www.gartner.com/en/newsroom/press-releases/2025-09-30-gartner-survey-finds-just-15-percent-of-it-application-leaders-are-considering-piloting-or-deploying-fully-autonomous-ai-agents
[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aN1QEwvpKGU-r-lMPx7Q8QAAA1U&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aN1QEwvpKGU-r-lMPx7Q8QAAA1U&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aN1QEwvpKGU-r-lMPx7Q8QAAA1U&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aN1QEwvpKGU-r-lMPx7Q8QAAA1U&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[6] https://www.theregister.com/2025/06/29/ai_agents_fail_a_lot/
[7] https://fortune.com/2025/05/09/klarna-ai-humans-return-on-investment/
[8] https://www.reddit.com/r/Futurology/comments/1kuilps/duolingo_ceo_walks_back_aifirst_comments_i_do_not/
[9] https://www.theregister.com/2024/08/29/ai_hiring_freeze_klarna/
[10] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aN1QEwvpKGU-r-lMPx7Q8QAAA1U&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[11] https://www.theregister.com/2025/09/02/salesforce_4000_jobs_ai/
[12] https://www.theregister.com/2025/06/16/salesforce_llm_agents_benchmark/
[13] https://www.theregister.com/2025/09/24/citi_pilots_agentic_ai/
[14] https://www.theregister.com/2025/09/23/developers_genai_little_productivity_gains/
[15] https://www.theregister.com/2025/09/17/gartner_ai_spending/
[16] https://www.theregister.com/2025/09/12/gartner_support_ai/
[17] https://www.theregister.com/2025/06/16/bt_chief_says_ai_could_cut_more_staff/
[18] https://www.theregister.com/2023/05/18/bt_redundancies_2030/
[19] https://newsroom.lufthansagroup.com/en/lufthansa-group-is-consistently-pursuing-its-strategy-and-aims-to-significantly-increase-profitability/
[20] https://www.ft.com/content/c9f905a0-cbfc-4a0a-ac4f-0d68d0fc64aa
[21] https://www.mckinsey.com/capabilities/quantumblack/our-insights/one-year-of-agentic-ai-six-lessons-from-the-people-doing-the-work
[22] https://whitepapers.theregister.com/
Consistency matters it turns out
breakfast
I understand it's core to how LLMs work, but it seems truly bizarre to take the principle of "do it differently every time" into your agentic systems. Most tasks need to be done the same way most of the time and a good worker will home in on a reliable approach and use it consistently. Even in the lowliest administrative role if somebody approached their tasks in a random and different way every time they'd be out the door in days, but this seems to be the standard approach for agentic systems.
The consequence is that nobody who takes their business seriously can use them for any task where it matters that it gets done.
Just 7%
Joe Gurman
Seems like a lot, really. I mean it took only one outfit, Cyberdyne, to end civilization as we know it by deploying autonomous AI.
Enterprises aren't keen on letting autonomous agents take the wheel…
No shit? Really?
Gartner needed to do a survey to confirm this? The mind boggles