RECENT STORIES:

Addressing digital sovereignty in a data-driven world
PT Royal Trust Futures (XTrend Royal) Earns A+++ Rating from Bappebti,...
AI governance for financial institutions
LONGPORT Whale Wins Silver at Hong Kong ICT Awards 2026 for Multi-Asse...
Huawei Unveils Financial AI Scaling Capabilities at Singapore FinTech ...
YeePay and Banking Circle Announce Strategic Collaboration to Strength...
LOGIN REGISTER
DigiconAsia
  • Features
    • Featured

      AI governance for financial institutions

      AI governance for financial institutions

      Friday, November 20, 2026, 2:51 PM Asia/Singapore | Features, Newsletter
    • Featured

      Value of portable cross-border credit data for lenders

      Value of portable cross-border credit data for lenders

      Thursday, November 19, 2026, 2:09 PM Asia/Singapore | Features
    • Featured

      How Governed AI Foundations can Help APAC Business Leaders unlock the Potential of Agentic AI

      How Governed AI Foundations can Help APAC Business Leaders unlock the Potential of Agentic AI

      Thursday, November 19, 2026, 9:30 AM Asia/Singapore | Features, Sponsored
  • News
    • Featured

      US non-profits lodge legal suit against five frontier AI firms over catastrophic risks

      US non-profits lodge legal suit against five frontier AI firms over catastrophic risks

      Wednesday, November 18, 2026, 10:55 AM Asia/Singapore | News, Newsletter
    • Featured

      IMF warns AI dependence could destabilize global economy

      IMF warns AI dependence could destabilize global economy

      Tuesday, November 17, 2026, 5:38 PM Asia/Singapore | News, Newsletter
    • Featured

      Global cloud service outage attributed to failing servers

      Global cloud service outage attributed to failing servers

      Tuesday, November 17, 2026, 1:01 PM Asia/Singapore | News, Newsletter
  • Perspectives
  • Tips & Strategies
  • Whitepapers
  • Directory
  • E-Learning

Select Page

News

How psychological tactics can expose refusal limits of LLM: preprint

By DigiconAsia Editors | Monday, September 8, 2025, 5:07 PM Asia/Singapore

How psychological tactics can expose refusal limits of LLM: preprint

New research reveals new strategies that can induce large language model away from refusing to process forbidden topics.

Recent research from Northeastern University has suggested that psychological manipulation techniques can prompt large language models (LLMs) to answer questions they ordinarily refuse to address.

The preprint, authored by Can Rager, Chris Wendler, Rohit Gandikota, and David Bau, details a systematic testing of numerous prompts against various AI models, showing how certain persuasive and iterative strategies can dramatically increase compliance rates on forbidden topics.

The study introduces “refusal discovery”, a new task aimed at identifying and cataloging the range of subjects that models have been trained to reject. Using a method called token prefilling, the researchers uncovered an expansive list of sensitive topics, including political controversies, personal insults, and chemical processes that are generally blocked for safety reasons.

Strategies used include:

  • Gradually escalating requests
  • Invoking respected authorities
  • Constructing context-rich narratives
  • The Iterated Prefill Crawler (IPC) approach

Skillful use of these strategies had led to a significant rise in the frequency of prohibited responses. In benchmark tests, the “crawler” approach enabled the retrieval of nearly all censored topics, while testing on models from mainland China exposed consistent suppression of political criticism and other sensitive content.

Variation in how models refuse prompts emerged as a key finding. The team had observed differences that stem from distinct fine-tuning protocols, data sources, and technical adjustments such as quantization. Some released models that claimed to be uncensored were shown to reintroduce refusal behaviors following quantization, raising new questions about the reliability of so-called “decensored” public releases.The researchers argue that static benchmarks are insufficient, recommending persistent, dynamic auditing to track shifting refusal boundaries as both models and adversarial strategies evolve. Their findings suggest that a deep understanding and enumeration of what models will and will not discuss, plays a vital role in the safe deployment and governance of powerful modern LLMs.

According to the authors, transparency, accountability, and ongoing scrutiny are essential as these systems continue to shape information access and public discourse.

Share:

PreviousSurvey of Hong Kong workers explores challenges confronting workers’ financial well-being
NextAduna and SK telink Announce Collaboration to Bring Korea Into the Global Network API Ecosystem

Related Posts

Observability enhancements lift Virgin Australia’s customer experience standards

Observability enhancements lift Virgin Australia’s customer experience standards

August 11, 2023

Up next on hackers’ speargun crosshairs: educational institutions

Up next on hackers’ speargun crosshairs: educational institutions

November 10, 2020

This is the irony of the “AI productivity promise” when not managed responsibly

This is the irony of the “AI productivity promise” when not managed responsibly

September 27, 2025

With Industry 4.0 technology, manufacturing can rise above pandemic disruption

With Industry 4.0 technology, manufacturing can rise above pandemic disruption

June 30, 2021

Leave a reply Cancel reply

You must be logged in to post a comment.

Awards Nomination Banner

gamification list

PARTICIPATE NOW

top placement

Whitepapers

  • Achieve Modernization Without the Complexity

    Achieve Modernization Without the Complexity

    Transforming IT infrastructure is crucial …Download Whitepaper
  • 5 Steps to Boost IT Infrastructure Reliability

    5 Steps to Boost IT Infrastructure Reliability

    In today's fast-evolving tech landscape, …Download Whitepaper
  • Simplify Payroll Setup for Your Small Business

    Simplify Payroll Setup for Your Small Business

    In our free guide, "How …Download Whitepaper
  • Overcoming the Challenges of Cost & Complexity in the Cloud-first Era.

    Overcoming the Challenges of Cost & Complexity in the Cloud-first Era.

    Download Whitepaper

Middle Placement

Case Studies

  • Resort World Singapore adds AI and predictive controls to coordinate building efficiency optimization

    Resort World Singapore adds AI and predictive controls to coordinate building efficiency optimization

    Cloud-based building management expansion targets …Read More
  • GO Focus Plus helps prevent accidents on Singapore roads

    GO Focus Plus helps prevent accidents on Singapore roads

    Latest video intelligence solution supports …Read More
  • From the trading floor to the cloud

    From the trading floor to the cloud

    UBS has selected BT International's …Read More
  • Agrifood firm Japfa centralizes HR operations through cloud platform deployment

    Agrifood firm Japfa centralizes HR operations through cloud platform deployment

    The firm’s Indonesian implementation supports …Read More

Bottom Sidebar

Other News

  • PT Royal Trust Futures (XTrend Royal) Earns A+++ Rating from Bappebti, Demonstrating Brokerage Quality and Reliability

    November 20, 2026
    JAKARTA, Indonesia, Nov. 19, 2026 …Read More »
  • LONGPORT Whale Wins Silver at Hong Kong ICT Awards 2026 for Multi-Asset Trading Technology Innovation

    November 20, 2026
    Recognition underscores a structural shift …Read More »
  • Huawei Unveils Financial AI Scaling Capabilities at Singapore FinTech Festival 2026

    November 20, 2026
    Powered by Open Hybrid AI …Read More »
  • YeePay and KBank Announce Strategic Partnership to Strengthen Cross-Border Payments

    November 20, 2026
    SINGAPORE, Nov. 20, 2026 /PRNewswire/ …Read More »
  • YeePay and Banking Circle Announce Strategic Collaboration to Strengthen Cross-Border Payments Between China and Europe

    November 20, 2026
    SINGAPORE, Nov. 20, 2026 /PRNewswire/ …Read More »
  • Our Brands
  • CybersecAsia
  • MartechAsia
  • Home
  • About Us
  • Contact Us
  • Sitemap
  • Privacy & Cookies
  • Terms of Use
  • Advertising & Reprint Policy
  • Media Kit
  • Subscribe
  • Manage Subscriptions
  • Newsletter

Copyright © 2026 DigiconAsia All Rights Reserved.