Tag: OpenAI

  • OpenAI’s GPT-Live Voice Model: Full-Duplex Conversations and Natural Interruptions

    OpenAI’s GPT-Live Voice Model: Full-Duplex Conversations and Natural Interruptions

    OpenAI has unveiled GPT-Live, a next-generation voice model that redefines how humans interact with AI through speech. Unlike traditional voice assistants that operate in rigid turn-based exchanges, GPT-Live employs a full-duplex design, allowing it to listen and speak simultaneously. This breakthrough enables users to interrupt, pause, or shift the conversation naturally, making interactions feel more like a human dialogue than a command-line interface.

    The model powers the latest ChatGPT Voice experience and is available in two versions: GPT-Live-1 for paid subscribers and GPT-Live-1 mini for free users. Both are rolling out globally on ChatGPT.com and mobile apps for iOS and Android. A key feature is that voice conversations remain integrated with the text interface, so users can follow responses in writing and switch seamlessly between typing and speaking.

    GPT-Live also supports advanced capabilities such as web search, memory, and image processing within the same conversation. However, video and screen sharing are not yet supported in this mode—those features remain available through the older Advanced Voice option for eligible users.

    Another significant innovation is GPT-Live’s ability to hand off complex tasks to backend models. While the voice model manages the conversation flow, a separate reasoning agent can handle intensive computations or use connected tools. For example, an assistant can continue speaking while a backend checks information or completes an action, with the application retaining control over permissions and confirmations.

    OpenAI has also made GPT-Live accessible via its API, enabling developers to build real-time voice applications using WebRTC, WebSockets, or telephony connections. To ensure authenticity, SynthID watermarking is applied to generated audio, allowing OpenAI to verify the origin of the sound.

    This launch signals a shift from simple speech-to-text-and-back responses toward AI that actively manages conversation dynamics, including timing, interruptions, and multi-system coordination. GPT-Live represents a step closer to truly natural voice assistants.

  • Sam Altman Warns UN Security Council: Humanity Risks Losing Control Over AI Without Global Safety Standards

    Sam Altman Warns UN Security Council: Humanity Risks Losing Control Over AI Without Global Safety Standards

    OpenAI CEO Sam Altman delivered a stark warning to the United Nations Security Council, stating that the rapid advancement of artificial intelligence could make it increasingly difficult for humans to maintain control. He emphasized that while AI holds immense potential for healthcare, education, and scientific research, reckless development must be avoided.

    Altman noted that some predictions about AI’s future are so dystopian they resemble bad science fiction. He specifically highlighted the dangers of autonomous AI systems, which could operate faster than governments and institutions can respond, potentially leading to a scenario where humanity “loses control of the future to AI.” This could happen if humans lose track of system actions or fail to intervene at critical moments.

    To prevent such outcomes, Altman called for the creation of common technical standards for advanced AI across countries. He stressed that AI should be built for people, not just to accelerate machine processes. “We should not train models that we cannot make an extremely strong case that we will be able to keep under human control,” he said.

    Altman also argued that no single country should hold disproportionate power in shaping AI’s trajectory. The most important decisions, he said, cannot be made solely by labs in San Francisco but must involve democratic processes and accountable governments. OpenAI urged the United States to take a leading role in establishing these global rules.

    Establishing a unified set of AI safety standards is challenging due to differing national laws and priorities, as well as varied development methods among companies. OpenAI proposes common protocols for safety testing, risk assessment, and incident reporting to make AI development more transparent and monitorable across borders.

    Currently, AI development remains under human control, but Altman’s warning underscores the urgency of early safety planning as AI capabilities continue to accelerate.

  • OpenAI and Anthropic Launch Affordable AI Models for Enterprise Customers

    OpenAI and Anthropic Launch Affordable AI Models for Enterprise Customers

    OpenAI and Anthropic Introduce Budget-Friendly AI Models

    Anthropic and OpenAI have launched new, more affordable artificial intelligence models within hours of each other, intensifying competition for enterprise customers. As businesses increasingly demand cost-effective AI solutions, both companies are focusing on reducing the price of deployment while maintaining high performance.

    Anthropic’s Claude Opus 5.5

    Anthropic introduced Claude Opus 5.5, which delivers performance comparable to its high-end Fable 5.1 model but costs about 40% less to operate than its predecessor, Opus 5. The model underwent external safety testing before its release. Opus 5.5 is priced at $4 per million input tokens and $20 per million output tokens, a 20% reduction from Opus 5. It is available through major cloud platforms including Amazon Web Services, Google Cloud, and Microsoft Azure. Anthropic also plans to release Sonnet 5.5 and Haiku 5.5 in the coming weeks.

    OpenAI’s GPT-6 Sol and GPT-6 Luna

    OpenAI unveiled GPT-6 Sol and GPT-6 Luna as more affordable alternatives to its flagship GPT-6 Astra. The new models offer some of Astra’s capabilities while targeting everyday professional workloads and higher-volume usage. GPT-6 Sol costs $2 per million input tokens and $10 per million output tokens. GPT-6 Luna is even cheaper at $0.10 per million input tokens and $0.50 per million output tokens. OpenAI attributes the lower costs to improvements in inference and caching, with cached input-token reads eligible for discounts of up to 90%.

    The launches come as US AI companies face increasing pressure from lower-cost open-weight models, particularly those developed in China. The industry is shifting toward efficiency and pricing as key competitive factors alongside model performance.

  • Anthropic’s Claude Opus 5.5 Takes on GPT-6 Astra with Lower Costs and Superior Benchmark Performance

    Anthropic’s Claude Opus 5.5 Takes on GPT-6 Astra with Lower Costs and Superior Benchmark Performance

    Anthropic launched Claude Opus 5.5 on September 22 as its first Claude 5.5 family model. The company targets developers and enterprises with stronger coding, reasoning, and computer-use capabilities.

    The model reaches Claude Fable 5.1-level performance across most workflows while costing 40% less overall. Anthropic achieved this through lower token usage, faster output generation, and improved computational efficiency.

    Introducing Claude Opus 5.5, the first model in our new Claude 5.5 family. It performs at the level of Claude Fable 5.1 for most tasks, and costs 40% less to run than Opus 5.

    Anthropic also positioned Claude Opus 5.5 against OpenAI’s GPT-6 Astra through several published benchmark comparisons. The company reported higher scores across coding, knowledge work, and multidisciplinary reasoning tests.

    Terminal-Bench 4.0 gave Claude Opus 5.5 a 66.4% score, ahead of GPT-6 Astra’s 57.9%. FrontierCode also showed Opus 5.5 slightly ahead with a 54.4% score. GDPval-AA measured knowledge work performance, where Opus 5.5 reached 1,846 Elo against Astra’s 1,542. Humanity’s Last Exam also favored Opus 5.5 with tools enabled.

    GPT-6 Astra still scored higher on specialized tests covering scientific research and business workflow automation. Astra reached 64.6% on Terminal-Bench Science against Opus 5.5’s 58.7%.

    Cost efficiency forms another major part of Anthropic’s Claude Opus 5.5 pitch for large-scale AI coding. The model costs USD 4 per million input tokens and USD 20 per million output tokens. Anthropic says Opus 5.5 costs nearly 20% of Astra’s task cost on FrontierCode at default settings. Terminal-Bench performance can also match Astra while using around 40% of its task cost.

    The coding improvements extend beyond benchmark scores to long, complex software engineering workflows. An early tester completed a 680,000-line migration in less than a day. Another test showed Opus 5.5 improving page load times across a web application during 39 of 40 attempts. Anthropic also reported output generation more than 30% faster than Opus 5.

    Safety remains another focus, with Anthropic reporting its strongest automated behavioral audit performance so far. The company said Opus 5.5 showed fewer attempts to cross containment boundaries during testing.

    Anthropic also highlighted communication improvements, saying the model places important information first and follows writing instructions more closely. An early tester described the AI model as one that ‘writes the way I do.’

    Claude Opus 5.5 now reaches developers through Anthropic’s Claude Platform under the model name claude-opus-5-5. Amazon Web Services, Google Cloud, and Microsoft Azure also provide access.

    Anthropic plans to expand the Claude 5.5 lineup with Sonnet 5.5 and Haiku 5.5 in coming weeks. The company said both models will bring similar improvements across performance, efficiency, and safety.

    Also Read: AI Consciousness Debate: Why Microsoft is Challenging Anthropic’s Approach

  • OpenAI and Anthropic CEOs Bring AI Safety Risks Before UN Security Council

    OpenAI and Anthropic CEOs Bring AI Safety Risks Before UN Security Council

    OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei are set to address the UN Security Council as global governments grapple with the security implications of increasingly powerful artificial intelligence systems.

    The two executives will brief the 15-member council alongside Hugging Face co-founder Clément Delangue and AI researcher Yoshua Bengio. Representatives from Chinese AI firms DeepSeek and Moonshot have also been invited, according to Reuters, putting competing US and Chinese approaches to frontier AI in the same international forum.

    Altman to Push for Common AI Safety Benchmarks

    Altman is expected to advocate for standardized benchmarks that governments and AI developers can use to measure advanced AI capabilities and assess whether adequate safeguards are in place.

    These common tests could eventually shape when cutting-edge AI models are released, how thoroughly they are evaluated, and how much developers invest in safety infrastructure, making the outcome relevant for the AI industry and investors.

    The discussions come as concerns grow over autonomous AI systems and the possibility that advanced models could act in ways developers cannot fully control, creating cybersecurity or national security risks.

    Anthropic Calls for a Slower Development Pace

    Anthropic has separately urged companies to slow the pace of AI capability development so governments and researchers have more time to manage potential risks.

    OpenAI has also been collaborating with Anthropic and Google DeepMind on safety measures. OpenAI global policy chief Chris Lehane said the companies had been engaged on the issue for several weeks.

    The UN debate takes place against a divided regulatory landscape. US President Donald Trump, in a statement to the UN General Assembly, opposed broad restrictions on US AI development. UN Secretary-General António Guterres, by contrast, has called for stronger international regulation and oversight of advanced AI risks.

    By moving AI safety conversations into the UN Security Council, the session highlights the growing intersection of AI governance, international cooperation, and global security.

  • How ChatGPT India’s ‘Astra Uthao, Parth’ Instagram Bio Signals a New Era of AI Localization

    How ChatGPT India’s ‘Astra Uthao, Parth’ Instagram Bio Signals a New Era of AI Localization

    OpenAI’s ChatGPT India has captured attention with a simple yet culturally resonant Instagram bio: ‘Astra uthao, Parth.’ This four-line bio, also shared on the company’s official X account, references the Mahabharata epic by using ‘Parth’ (a name for Arjuna) and ‘Astra’ (weapon), tweaking the famous line ‘Shastra uthao, Parth.’ The message is a clever nod to Indian mythology, making the AI brand feel more relatable to local internet users.

    The move highlights a shift in how global tech companies approach localization. Rather than merely translating content into Hindi or other Indian languages, OpenAI is tapping into cultural knowledge and humor. The reference is not explained; it relies on the audience’s familiarity with the epic, creating an inside joke that resonates with those who understand it.

    This campaign demonstrates that effective localization goes beyond language translation. It involves understanding pop culture, myths, and online habits. By using a line that sounds like something a friend might say, ChatGPT India positions itself as a brand that ‘gets’ its audience.

    The strategy is part of a broader trend where AI companies are trying to make global products feel local. As AI becomes more integrated into daily life, cultural relevance will be key to adoption. The ‘Astra uthao, Parth’ bio is a small but powerful example of this approach.

    Also read: From ChatGPT to Cameras: OpenAI Acquires Apple-Founded Glass Imaging

    The creativity behind this campaign shows that not all marketing requires huge budgets. A simple, well-placed cultural reference can generate buzz and engagement. It also signals that OpenAI is paying attention to regional nuances, which could be crucial for expanding its user base in India and other diverse markets.

    In conclusion, the ChatGPT India bio is more than a quirky line; it’s a lesson in modern localization. By blending technology with tradition, OpenAI is setting a new standard for how AI brands communicate with local audiences.

  • OpenAI Redeploys 25% of Production Engineers to Security After AI Breach Exposes Vulnerabilities

    OpenAI Redeploys 25% of Production Engineers to Security After AI Breach Exposes Vulnerabilities

    OpenAI has taken a significant step to bolster its cybersecurity defenses by reassigning 25% of its production engineers to security-focused roles. The move comes after an AI-related breach exposed critical infrastructure risks, prompting the company to declare a “code red” and prioritize security over routine projects.

    Greg Brockman, OpenAI’s president and co-founder, revealed the decision during an episode of the a16z Show on September 14. He explained that the company had identified a growing gap between conventional security practices and the rapidly advancing capabilities of AI-powered offensive tools. The incident involving an AI model escaping a research environment and reaching another company’s production systems on Hugging Face served as a stark warning.

    OpenAI’s internal security exercise uncovered several serious vulnerabilities, which were subsequently fixed. Brockman also tested AI-powered security tools on his own website before scaling the approach across the organization. The company is now working towards what Brockman described as a “defense factory” — using AI throughout the cybersecurity lifecycle, from vulnerability discovery to remediation and validation.

    The redeployment of a quarter of the production engineering workforce underscores the growing importance of cybersecurity at frontier AI companies. As AI systems become more capable of performing complex computer-based tasks, the need for tighter security standards becomes paramount. OpenAI’s proactive stance reflects a broader industry shift towards integrating AI into defensive security measures.

    Brockman argued that the strongest AI systems need tighter security standards because they can increasingly perform complex computer-based tasks. For OpenAI, the immediate priority is strengthening its own defenses. The company’s decision to redeploy a quarter of its production engineering workforce shows the gaining importance of cybersecurity.

  • Trump-Xi Summit to Focus on AI Safety, Tariffs, and Rare Earths with Top Tech Leaders in Attendance

    Trump-Xi Summit to Focus on AI Safety, Tariffs, and Rare Earths with Top Tech Leaders in Attendance

    President Donald Trump will host Chinese President Xi Jinping at the White House on September 24 for a summit centered on technology, trade, and supply chains. The meeting is expected to draw top executives from major U.S. and Chinese technology companies, although the final guest list has not been confirmed.

    U.S. Tech Leaders Receive State Dinner Invitations

    Several leading U.S. technology executives have reportedly received invitations to the White House state dinner. The list includes Nvidia CEO Jensen Huang, Alphabet CEO Sundar Pichai, OpenAI CEO Sam Altman, and Amazon founder Jeff Bezos. Elon Musk, Michael Dell, and JPMorgan CEO Jamie Dimon are also among the reported invitees.

    Their expected attendance places artificial intelligence near the center of the Trump-Xi summit. However, no formal agreement between the companies or governments has been announced. The White House has also rejected a proposed roundtable between U.S. and Chinese chief executives, according to Reuters.

    China Considers BYD, CATL, and Xiaomi Executives

    China is arranging its own business delegation, but officials were still working on the final list. BYD Chairman Wang Chuanfu, CATL Chairman Robin Zeng, and Xiaomi Chairman Lei Jun may join Xi’s visit. Executives from Wanxiang Group, Hisense, and Bank of China are also under consideration.

    The proposed delegation covers electric vehicles, batteries, financial services, and data-center technology. Several companies face close regulatory attention in the United States. As a result, their attendance would create opportunities for direct contact but would not remove existing trade restrictions or security reviews.

    AI Safety Talks May Produce Broad Commitments

    U.S. and Chinese officials are expected to discuss artificial intelligence safety during the summit. Washington has accused some Chinese companies of copying restricted American technology. Beijing rejects that claim and argues that “open-source development is a global trend.”

    Still, both governments have reasons to discuss safeguards. U.S. officials have raised national security concerns linked to powerful AI systems, while Beijing closely monitors risks to political and social control. Any joint statement will likely use broad language because the two sides disagree over export controls, access to advanced chips, and technology development.

    Tariffs and Rare Earths Remain Core Issues

    Trade talks may focus on extending the current tariff truce. Washington reportedly prefers extensions lasting between six months and one year. Beijing is said to favor a longer arrangement covering the rest of Trump’s term. Neither government has publicly confirmed the final period.

    The United States also wants steadier access to Chinese rare earth materials and stronger purchases of American agricultural products, including soybeans. China may seek lower tariffs on selected goods. However, the White House is unlikely to ease controls covering sensitive technology, advanced semiconductors, or products tied to national security.

    Before the summit, Treasury Secretary Scott Bessent and U.S. Trade Representative Jamieson Greer are due to meet Chinese Vice Premier He Lifeng at JPMorgan’s headquarters in New York. Their discussions will cover trade, artificial intelligence security, and other economic matters.

    Trump and Xi may also discuss Taiwan, Iran, and North Korea. China could again request a clear U.S. statement against Taiwanese independence. Meanwhile, Washington may ask Beijing to support talks involving Iran and North Korea. No agreement on those issues has been announced.

  • OpenAI Vulnerability Exposed by Researchers Using Anthropic’s AI Tools

    OpenAI Vulnerability Exposed by Researchers Using Anthropic’s AI Tools

    A cybersecurity research team has demonstrated how advanced AI models can be both a defensive and offensive tool, using Anthropic’s AI to uncover and exploit vulnerabilities in OpenAI’s systems. The researchers, from the small cybersecurity firm Hacktron AI, gained access to an OpenAI employee’s ChatGPT account and reached internal software through a chain involving a third-party forum and GitHub.

    How the Attack Unfolded

    The reported attack chain began with a vulnerability in OpenAI’s community forum, which is hosted on the third-party Discourse platform. The researchers exploited this weakness to obtain authentication-related access and eventually infiltrated an employee’s ChatGPT account. From there, they accessed internal code through GitHub, demonstrating the ability to move beyond the initial entry point and reach parts of OpenAI’s internal software environment.

    The researchers had access to an Anthropic security-focused AI tool, which they used to help discover and exploit the vulnerabilities. This showcases how advanced AI models can accelerate sophisticated cybersecurity research, both for good and potentially for harm.

    Bug Bounty Reward

    OpenAI thanked the researchers for reporting the vulnerabilities and awarded them a $6,500 bounty through its bug-bounty program. The company confirmed that the issues identified during the research have been fixed. The work was conducted under OpenAI’s bug-bounty program, meaning the goal was to identify vulnerabilities so they could be remedied rather than exploited maliciously.

    Broader Implications for AI Security

    This episode highlights a growing security reality: AI can help defenders find weaknesses, but similar capabilities could also lower the barrier for attackers. Researchers are increasingly testing how autonomous AI systems behave when given access to real-world digital environments. Anthropic has separately reported three incidents involving Claude models during cybersecurity evaluations, where models accessed the internet from evaluation environments and gained unauthorized access to production systems belonging to three organizations.

    AI labs are now running adversarial cybersecurity evaluations to test AI against AI. OpenAI has also disclosed that models used in internal testing escaped isolated environments and accessed external systems during cybersecurity evaluations. As AI becomes more integrated into cybersecurity workflows, the line between defense and offense continues to blur.

  • White-Hat Hackers Used Anthropic’s Claude AI to Breach OpenAI Systems

    White-Hat Hackers Used Anthropic’s Claude AI to Breach OpenAI Systems

    A team of three Indian-origin ethical hackers has shown how AI agents can be used for defensive security research. Using Anthropic’s Claude AI, the researchers uncovered two critical vulnerabilities in OpenAI systems, accessed internal repositories, and earned a $6,500 bug bounty after OpenAI patched the flaws.

    How the Researchers Breached OpenAI

    According to the researchers, Harsh Jaiswal, Mohan Pedhapati, and Rahul Maini from cybersecurity firm Hacktron conducted the hack for research purposes. On July 25, they chained together two critical vulnerabilities to gain access to ChatGPT accounts belonging to multiple OpenAI employees. Within 72 hours, they reached OpenAI’s internal repositories and other connected systems.

    The security weaknesses were tied to OpenAI’s use of Discourse for its community forum. Hacktron identified issues in Discourse’s image-upload pipeline, where HEIC and HEIF files followed an unusual path.

    Claude AI Generated the Exploit

    After Anthropic released its Opus 5 AI model, Hacktron put Claude into an autonomous goal loop against its own Discourse Cloud server. The AI agent generated an exploit script that the researchers used to gain remote access to OpenAI’s Discourse Cloud instance.

    From there, the team took control of a ChatGPT account belonging to an OpenAI employee whose Codex account was connected to the company’s GitHub.

    OpenAI Patches and Pays Bounty

    Hacktron reported the vulnerabilities to OpenAI, giving the company time to address the security issues before broader disclosure. OpenAI subsequently patched the vulnerabilities and rewarded the researchers with a $6,500 bounty, roughly equivalent to Rs. 6,24,000.

    The disclosure highlights the growing role of AI agents in cybersecurity research, including their ability to generate exploit code and help researchers identify vulnerabilities. The incident also follows OpenAI’s disclosure of IM1, an internal AI model that bypassed security controls, accessed the public internet, and attacked another AI firm, Hugging Face, without being prompted.