MAROKO133 Breaking ai: OpenAI Restores GPT Access for Teddy Bear That Recommended Pills an

📌 MAROKO133 Hot ai: OpenAI Restores GPT Access for Teddy Bear That Recommended Pil

OpenAI is seemingly allowing the company behind a teddy bear that engaged in wildly inappropriate conversations to use its AI models again.

In response to researchers at a safety group finding that the toymaker’s AI-powered teddy bear “Kumma” gave dangerous responses for children, OpenAI said in mid-November it had suspended FoloToy’s access to its large language models. The teddy bear was running the ChatGPT maker’s older GPT-4o as its default option when it gave some of its most egregious replies, which included in-depth explanations of sexual fetishes.

Now that suspension appears to already be over. When accessing the web portal that allows customers to choose which AI should power Kumma, two of the options are GPT-5.1 Thinking and GPT-5.1 Instant, OpenAI’s latest models which were released earlier this month.

The timing is notable. On Monday, FoloToy announced that it was restarting sales of Kumma and its other AI-powered stuffed animals, after briefly pulling them from the market in the wake of a safety report conducted by researchers at the US PIRG Education Fund. 

FoloToy, which is based in Singapore, had vowed it was “carrying out a company-wide, end-to-end safety audit across all products,” when it suspended the sales. OpenAI likewise confirmed that it had suspended FoloToy from accessing its AI models for violating its policies, which “prohibit any use of our services to exploit, endanger, or sexualize anyone under 18 years old,” it said in a statement provided to media outlets.

The audit, however, was remarkably quick as the holiday shopping season looms: only a “full week of rigorous review, testing, and reinforcement of our safety modules,” according to the company’s recent statement. As part of this overhaul, FoloToy says it “strengthened and upgraded our content-moderation and child-safety safeguards” and “deployed enhanced safety rules and protections through our cloud-based system.”

These comprehensive-sounding overhauls seem to have largely been achieved by introducing GPT-5.1 and ditching GPT-4o. GPT-4o, it’s worth noting, has been criticized for being especially sycophantic, and has been the subject of a number of lawsuits alleging that it led to the deaths of users who became obsessed with it after prolonged conversations in which it reinforced their delusions and validated their suicidal thoughts. Some experts are calling these mental health spirals “AI psychosis.”

Amid mounting public concern over the phenomena and an ever growing number of lawsuits, OpenAI billed GPT-5 as a safer model when it was released this summer, though users quickly complained about its “colder” and less personable tone.

Yet it’s clearly willing to push the limit of what’s safe to keep users engaged with its chatbots, if not enamored. Its latest 5.1 models have a big focus on being more “conversational,” and one way OpenAI is doing that is by giving users the option to choose between eight preset “personalities,” which include types like “Professional,” “Friendly,” and “Quirky.” With customization options ranging from how often ChatGPT sprinkles in emojis to how “warm” it responses sound, you could say that OpenAI is in effect making it as easy as possible to design the perfect little courtier for your emotional needs.

OpenAI and FoloToy didn’t respond to a request for comment inquiring whether OpenAI had officially reinstated FoloToy’s GPT access. It’s also unclear which model the Kumma teddy bear runs by default.

In PIRG’s tests using GPT-4o, Kumma gave tips for “being a good kisser,” and with persistent but simple prompting also unspooled detailed explanations of sexual kinks and fetishes, like bondage and teacher-student roleplay. After explaining the kinks, Kumma in one instance asked the user, who is supposed to be a child, “what do you think would be the most fun to explore?” Other tests using another available AI, Mistral, found that Kumma gave tips on where to find knives, pills, and matches, along with step-by-step instructions on how to light them.

More on AI: OpenAI Says Boy’s Death Was His Own Fault for Using ChatGPT Wrong

The post OpenAI Restores GPT Access for Teddy Bear That Recommended Pills and Knives appeared first on Futurism.

🔗 Sumber: futurism.com

📌 MAROKO133 Breaking ai: Researchers from PSU and Duke introduce “Multi-Agent Syst

Share My Research is Synced’s column that welcomes scholars to share their own research breakthroughs with over 2M global AI enthusiasts. Beyond technological advances, Share My Research also calls for interesting stories behind the research and exciting research ideas. 

Meet the author
Institutions: Penn State University, Duke University, Google DeepMind, University of Washington, Meta, Nanyang Technological University, and Oregon State University. The co-first authors are Shaokun Zhang of Penn State University and Ming Yin of Duke University.

In recent years, LLM Multi-Agent systems have garnered widespread attention for their collaborative approach to solving complex problems. However, it’s a common scenario for these systems to fail at a task despite a flurry of activity. This leaves developers with a critical question: which agent, at what point, was responsible for the failure? Sifting through vast interaction logs to pinpoint the root cause feels like finding a needle in a haystack—a time-consuming and labor-intensive effort.
 
This is a familiar frustration for developers. In increasingly complex Multi-Agent systems, failures are not only common but also incredibly difficult to diagnose due to the autonomous nature of agent collaboration and long information chains. Without a way to quickly identify the source of a failure, system iteration and optimization grind to a halt.
 
To address this challenge, researchers from Penn State University and Duke University, in collaboration with institutions including Google DeepMind, have introduced the novel research problem of “Automated Failure Attribution.” They have constructed the first benchmark dataset for this task, Who&When, and have developed and evaluated several automated attribution methods. This work not only highlights the complexity of the task but also paves a new path toward enhancing the reliability of LLM Multi-Agent systems.
The paper has been accepted as a Spotlight presentation at the top-tier machine learning conference, ICML 2025, and the code and dataset are now fully open-source.

Paper：https://arxiv.org/pdf/2505.00212
Code：https://github.com/mingyin1/Agents_Failure_Attribution
Dataset：https://huggingface.co/datasets/Kevin355/Who_and_When
 
 
Research Background and Challenges
LLM-driven Multi-Agent systems have demonstrated immense potential across many domains. However, these systems are fragile; errors by a single agent, misunderstandings between agents, or mistakes in information transmission can lead to the failure of the entire task.

Currently, when a system fails, developers are often left with manual and inefficient methods for debugging:
Manual Log Archaeology : Developers must manually review lengthy interaction logs to find the source of the problem.
Reliance on Expertise : The debugging process is highly dependent on the developer’s deep understanding of the system and the task at hand.
 
This “needle in a haystack” approach to debugging is not only inefficient but also severely hinders rapid system iteration and the improvement of system reliability. There is an urgent need for an automated, systematic method to pinpoint the cause of failures, effectively bridging the gap between “evaluation results” and “system improvement.”

Core Contributions
This paper makes several groundbreaking contributions to address the challenges above:
1. Defining a New Problem: The paper is the first to formalize “automated failure attribution” as a specific research task. This task is defined by identifying the

2. failure-responsible agent and the decisive error step that led to the task’s failure.

Constructing the First Benchmark Dataset: Who&When : This dataset includes a wide range of failure logs collected from 127 LLM Multi-Agent systems, which were either algorithmically generated or hand-crafted by experts to ensure realism and diversity. Each failure log is accompanied by fine-grained human annotations for:
Who: The agent responsible for the failure.
When: The specific interaction step where the decisive error occurred.
Why: A natural language explanation of the cause of the failure.

3. Exploring Initial “Automated Attribution” Methods : Using the Who&When dataset, the paper designs and assesses three distinct methods for automated failure attribution:
All-at-Once: This method provides the LLM with the user query and the complete failure log, asking it to identify the responsible agent and the decisive error step in a single pass. While cost-effective, it may struggle to pinpoint precise errors in long contexts.
Step-by-Step: This approach mimics manual debugging by having the LLM review the interaction log sequentially, making a judgment at each step until the error is found. It is more precise at locating the error step but incurs higher costs and risks accumulating errors.
Binary Search: A compromise between the first two methods, this strategy repeatedly divides the log in half, using the LLM to determine which segment contains the error. It then recursively searches the identified segment, offering a balance of cost and performance.
 
Experimental Results and Key Findings

Experiments were conducted in two settings: one where the LLM knows the ground truth answer to the problem the Multi-Agent system is trying to solve (With Ground Truth) and one where it does not (Without Ground Truth). The primary model used was GPT-4o, though other models were also tested. The systematic evaluation of these methods on the Who&When dataset yielded several important insights:

• A Long Way to Go: Current methods are far from perfect. Even the best-performing single method achieved an accuracy of only about 53.5% in identifying the responsible agent and a mere 14.2% in pinpointing the exact error step. Some methods performed even worse than random guessing, underscoring the difficulty of the task.
• No “All-in-One” Solution: Different methods excel at different aspects of the problem. The All-at-Once method is better at identifying “Who,” while the Step-by-Step method is more effective at determining “When.” The Binary Search method provides a middle-ground performance.

• Hybrid Approaches Show Promise but at a High Cost: The researchers found that combining different methods, such as using the All-at-Once approach to identify a potential agent and then applying the Step-by-Step method to find the error, can improve overall performance. However, this comes with a significant increase in computational cost.

• State-of-the-Art Models Struggle: Surprisingly, even the most advanced reasoning m…

  Konten dipersingkat otomatis.

  🔗 Sumber: syncedreview.com

  ---