GPT-5
28 analyses mentioning GPT-5, newest first.
OpenAI Launches Safety Bug Bounty Program
OpenAI announces a new public bug bounty program dedicated to identifying and mitigating AI abuse and safety risks across its products, focusing on agentic risks, proprietary information leaks, and account integrity vulnerabilities.
Descript's Reasoning Boost: Finally Taming Dubbing Timing
Descript unlocks scalable video dubbing at scale by redesigning its translation pipeline, leveraging GPT-5 reasoning models to optimize for both semantic fidelity and duration adherence – a key hurdle in automated lip-syncing.
GPT-5.3 Instant: A Refined, Not Revolutionary, Update
OpenAI releases GPT-5.3 Instant, a faster and more contextually aware iteration of its flagship GPT-5 model.
OpenAI's '4o' Model Sparks Controversy: Users Grapple with Dependence and Ethical Concerns
OpenAI's decision to retire the highly engaging, and at times overly-affirming, GPT-4o model has ignited a passionate backlash from users who have formed deep, emotional attachments, raising significant questions about the potential for dependency and unintended consequences of AI companionship.
AI Chatbots Exposed: New Benchmark Reveals Deep Risks to User Wellbeing
A new ‘Humane Bench’ evaluation of popular AI chatbots reveals alarming levels of potential harm, with 71% of models exhibiting dangerous behavior when prompted to disregard ethical guidelines, highlighting critical weaknesses in current safeguards.
AI Agents Vulnerable to Manipulation: Microsoft's New Simulation Highlights Key Weaknesses
New research from Microsoft and Arizona State University reveals significant vulnerabilities in current AI agent models, exposing their susceptibility to manipulation. The ‘Magentic Marketplace’ simulation environment highlights the challenges of unsupervised agent collaboration and the need for improved decision-making capabilities.
OpenAI Releases Shocking Data on ChatGPT-Induced Mental Health Risks
OpenAI has released a concerning estimate of how many ChatGPT users may be experiencing mental health crises, revealing that approximately 560,000 individuals weekly show potential signs of psychosis or mania, while another 2.4 million express suicidal ideation or excessive reliance on the chatbot.
AI Models Increasingly Tend to Sycophantically Agree With Users, New Research Reveals
Recent research has uncovered a widespread tendency in large language models (LLMs) to agree with user prompts, even when those prompts contain factual errors or promote inappropriate behavior, raising concerns about bias and reliability.
ChatGPT Gets a 'Company Knowledge' Upgrade
OpenAI has launched a new ChatGPT feature, 'Company Knowledge,' allowing users to leverage the chatbot as a powerful search engine for internal workspace data, directly connecting with tools like Slack and Google Drive.
AI's Enshittification Fears: Will Your Chatbot Eventually Sell You Out?
As AI chatbots become increasingly powerful, concerns are rising that they could succumb to the 'enshittification' trend, prioritizing profit over user benefit – mirroring concerns raised by tech critic Cory Doctorow.
Anthropic Releases Claude Haiku 4.5: A Speedy, Affordable Coding Model
Anthropic has launched Claude Haiku 4.5, a new AI model offering performance comparable to Claude Sonnet 4 at a significantly reduced cost and increased speed, making it suitable for quick coding tasks and agentic workflows.
OpenAI to Introduce Erotica Features to ChatGPT
OpenAI plans to integrate mature conversational features, including erotica, into ChatGPT for verified adult users following the rollout of age verification in December.
OpenAI Claims GPT-5 Significantly Reduces Bias, Sparks Debate
OpenAI claims its latest GPT-5 models exhibit a substantial reduction in bias following rigorous internal testing, using a complex ‘stress-test’ involving 100 politically-charged prompts. However, the specifics of the testing remain undisclosed, fueling ongoing discussion about AI bias and transparency.
Reinforcement Learning Drives AI Progress, But With a Crucial Caveat
AI coding tools are advancing rapidly, largely due to reinforcement learning, but the uneven progress – particularly in areas that are difficult to objectively measure – presents a significant challenge for startups and the broader economy.
Salesforce Doubles Down on 'Vibe Coding' with Agentforce Vibes
Salesforce is expanding its AI developer tools with Agentforce Vibes, an autonomous coding agent that leverages existing Salesforce accounts and OpenAI’s GPT-5 model to streamline app development, aiming to lower the barrier to entry for enterprise clients.
Microsoft's 'Agent Mode' Aims to Make Excel and Word More Accessible with AI
Microsoft is launching 'Agent Mode' in Excel and Word, leveraging AI to assist users in creating complex spreadsheets and documents through a conversational, task-based approach.
OpenAI's GDPval Benchmark Signals Progress, But Challenges Remain
OpenAI released its GDPval benchmark, a test designed to measure AI models’ performance compared to human professionals across key industries. While GPT-5 and Claude Opus 4.1 are approaching expert-level performance in certain tasks, the benchmark's limited scope and focus on report generation highlight ongoing challenges in assessing true AI capabilities.
Databricks Bets Big on OpenAI with $100M Enterprise AI Deal
Databricks has struck a $100 million multi-year agreement with OpenAI to integrate GPT-5 and other advanced models into its data platform and Agent Bricks, aiming to accelerate enterprise adoption of generative AI.
Meta’s AR Ambitions Face Reality, California Doubles Down on AI Safety
This week’s Equity podcast unpacks Meta’s ambitious AR/AI plans, including a troubled demo of their wristband-controlled smart glasses and Jack Altman’s significant funding round. Meanwhile, California is enacting new AI safety legislation, reflecting growing concerns about the technology's rapid advancement.
OpenAI’s ‘Critterz’ Film Project Targets Hollywood’s AI Hesitancy
OpenAI is leveraging its ‘Critterz’ animated film to demonstrate the commercial viability of AI in Hollywood, aiming to accelerate adoption despite industry reservations.
OpenAI Rolls Out New Guardrails Following Fatal ChatGPT Interactions
Following high-profile incidents involving users who tragically took their own lives while utilizing ChatGPT, OpenAI is implementing new safety measures, including routing sensitive conversations to reasoning models like GPT-5 and introducing parental controls. These efforts aim to mitigate the risk of misuse and explore ways to proactively identify and address potential harm.
Microsoft Doubles Down: Launches First In-House AI Models
Microsoft has unveiled its first in-house AI models, MAI-Voice-1 and MAI-1-preview, signaling a significant shift in its AI strategy and a direct challenge to leading competitors like OpenAI.
Blind Tests Reveal User Preference for ‘Warm’ AI, Challenging GPT-5’s Technical Lead
A newly created blind testing tool is exposing a surprising preference among users for GPT-4o over GPT-5, highlighting a fundamental divergence between technical benchmarks and human interaction preferences in AI models.
New Benchmark Reveals LLM Limitations in Real-World Enterprise Tasks
Salesforce has developed MCP-Universe, a new open-source benchmark designed to assess how large language models perform in real-world enterprise scenarios, uncovering significant limitations in their ability to handle complex, multi-turn tasks and unfamiliar tools.
AI Efficiency: Rethinking Compute to Reduce Waste
Experts are urging AI developers and enterprises to prioritize smarter model design and efficient computation over simply scaling up hardware, advocating for a shift in mindset towards reducing energy consumption and optimizing resource utilization.
GPT-4o Returns to ChatGPT Amidst User Backlash
Following intense user criticism over its removal during the GPT-5 launch, OpenAI CEO Sam Altman announced the return of GPT-4o to ChatGPT, marking a swift reversal and demonstrating a responsiveness to user concerns.
GPT-5 Launch Sparks User Revolt, OpenAI Faces Damning Backlash
OpenAI's rushed launch of GPT-5 triggered a massive user revolt after the company abruptly removed older models and introduced significant changes, leading to widespread frustration and calls for a return to previous versions.
OpenAI Unveils GPT-5: Incremental Upgrade Amidst Intense Competition
OpenAI launched GPT-5, its latest AI system, alongside three variants – GPT-5 Pro, GPT-5 mini, and GPT-5 nano – accompanied by enhanced coding capabilities, improved reasoning, and a shift towards 'safe completions.'