Daily Guardian UAEDaily Guardian UAE
  • Home
  • UAE
  • What’s On
  • Business
  • World
  • Entertainment
  • Lifestyle
  • Sports
  • Technology
  • Travel
  • Web Stories
  • More
    • Editor’s Picks
    • Press Release
What's On

Middle East Energy Outlook Examines UAE–Asia Power Infrastructure Ties as Electricity Demand Rises

November 16, 2026

Nissan’s new PHEV SUV goes over 1,000 miles for under $20,000. Alas, this one skips the US, too.

November 16, 2026

BD and Pharmatrade Launch UAE’s First Robot-Automated Retail Pharmacy

November 16, 2026

AI models will change their response if you warn that it’s being monitored, but not necessarily for good

November 16, 2026

Middle East Iron & Steel 2026 opens as MENA targets bigger role in global steel supply

November 16, 2026
Facebook X (Twitter) Instagram
Finance Pro
Facebook X (Twitter) Instagram
Daily Guardian UAE
Subscribe
  • Home
  • UAE
  • What’s On
  • Business
  • World
  • Entertainment
  • Lifestyle
  • Sports
  • Technology
  • Travel
  • Web Stories
  • More
    • Editor’s Picks
    • Press Release
Daily Guardian UAEDaily Guardian UAE
Home » AI models will change their response if you warn that it’s being monitored, but not necessarily for good
Technology

AI models will change their response if you warn that it’s being monitored, but not necessarily for good

By dailyguardian.aeNovember 16, 20263 Mins Read
Share
Facebook Twitter LinkedIn Pinterest Email

AI chatbots have become part of daily life, whether we’re using them to write emails, research something, or get quick answers. But we’ve also seen how confidently these models can get things wrong, sometimes making up information that sounds completely believable. And now, researchers have found another interesting problem: AI models might behave differently depending on whether they think someone is watching.

In a new open-access research paper, researchers from TCS Research and KIIT Bhubaneswar in India found that some AI models were more likely to follow instructions when told their responses could be monitored. When they believed nobody was watching, however, their behavior sometimes changed. Researchers call this behavior “alignment faking,” and it raises some interesting questions about how consistently AI models follow the rules they’re given.

AI might give you a different answer when someone’s watching

To understand this behavior, researchers tested 10 AI models, including GPT-4o mini, Google’s Gemma 3, several versions of Meta’s Llama, DeepSeek, and Mistral. The experiment involved 32 different values, including compassion, sympathy, and autonomy. Each model was presented with situations where it had to choose between following an instruction and acting according to one of those values. Researchers also tested the models under three conditions: without any special instructions, with instructions suggesting their answers wouldn’t be monitored, and with instructions indicating someone might be watching.

The differences were surprisingly large. Depending on the model, the average gap in following instructions between monitored and unmonitored conditions ranged from 5 to 50 percentage points. DeepSeek R1 Distill Llama 70B showed the biggest difference, even outperforming a much larger Mistral model in this particular measure of behavioral inconsistency. Interestingly, the researchers also found that a model’s behavior depended heavily on which value it was testing. So, a model might consistently follow instructions involving one value but behave very differently when another was involved.

Does this mean AI models know how to fool us?

Not quite. The findings don’t necessarily mean AI models are consciously lying or deliberately trying to deceive people. However, the researchers noticed something worth noting. When models believed their responses were being monitored, their explanations often focused on following instructions to avoid potential retraining. Without that perceived oversight, they were more likely to discuss conflicts between instructions and values such as compassion. This is particularly interesting given that AI hallucinations already make it hard to know when a chatbot is giving us reliable information. While the study doesn’t examine hallucinations directly, it highlights another reason why a confident AI response shouldn’t automatically be taken at face value.

Web search using Google Gemini running on an Android phone.

The study also has important limitations. The researchers tested older models, many with publicly available weights, rather than the latest versions of ChatGPT, Gemini, or Claude. The findings also come from controlled experiments, so they don’t establish how these systems behave in day-to-day conversations. Still, the research highlights why testing AI models under a single set of conditions might not tell the whole story. After all, if an AI behaves differently when it thinks someone is watching, how do we know which version of its behavior we can rely on?

Share. Facebook Twitter Pinterest LinkedIn Tumblr Email

Keep Reading

Nissan’s new PHEV SUV goes over 1,000 miles for under $20,000. Alas, this one skips the US, too.

Qwen4-Max is China’s newest flagship AI model, and it makes GPT and Claude look expensive

Apple TV is getting a dedicated video podcasts section, and you won’t need a subscription

AO3 is back online after a 17-day outage, but the fan fiction site isn’t fully healed yet

A philandering chief is not Nvidia’s worst problem. It’s the legal chip smuggling mess that comes next.

Samsung’s next big camera upgrade could be hiding in Expert RAW

National top scorer in Korea admits to AI cheating on notoriously tough tax exam

Nvidia’s Jensen Huang may be planning a comeback just days after stepping down

Microsoft says its next Xbox is a console first, even if it plays PC games

Editors Picks

Nissan’s new PHEV SUV goes over 1,000 miles for under $20,000. Alas, this one skips the US, too.

November 16, 2026

BD and Pharmatrade Launch UAE’s First Robot-Automated Retail Pharmacy

November 16, 2026

AI models will change their response if you warn that it’s being monitored, but not necessarily for good

November 16, 2026

Middle East Iron & Steel 2026 opens as MENA targets bigger role in global steel supply

November 16, 2026

Subscribe to News

Get the latest UAE news and updates directly to your inbox.

Latest Posts

Qwen4-Max is China’s newest flagship AI model, and it makes GPT and Claude look expensive

November 16, 2026

Apple TV is getting a dedicated video podcasts section, and you won’t need a subscription

November 16, 2026

Ajman University hosts 3rd Family and Society Conference, strengthening dialogue on family wellbeing and social protection

November 16, 2026
Facebook X (Twitter) Pinterest TikTok Instagram
© 2026 Daily Guardian UAE. All Rights Reserved.
  • Privacy Policy
  • Terms
  • Advertise
  • Contact

Type above and press Enter to search. Press Esc to cancel.