OpenAI has launched several new tools aimed at making it easier for developers to create applications powered by its AI technology. Among the key innovations is a real-time tool that allows developers to build AI voice applications using a single set of instructions, streamlining what was previously a multi-step process.
The startup, supported by Microsoft, also introduced a fine-tuning tool that enables developers to improve AI model responses using both text and images. This enhancement boosts capabilities like visual search and object detection, potentially benefiting sectors such as autonomous vehicles.
OpenAI has forecast a rapid rise in revenue, expecting to generate $11.6 billion next year, driven by businesses building their own AI apps using its technology. With competition from tech giants like Google heating up, OpenAI is focused on rolling out advanced tools to retain its edge in the generative AI race.
Other newly unveiled features include a method for smaller AI models to learn from larger ones, and a ‘Prompt Caching’ system that can reduce development costs by reusing previously processed text, cutting expenses by up to half.
As global investors like Thrive Capital and Tiger Global invested $6.6 billion in OpenAI, the company is seeking more than just capital; it wants assurances that these investors will avoid funding five perceived competitors. The list includes rivals such as Anthropic, Elon Musk’s xAI, and Safe Superintelligence (SSI), co-founded by OpenAI’s Ilya Sutskever. These companies are in a race to develop large language models, which require substantial financial backing.
OpenAI is also focusing on AI applications, with firms like the search startup Perplexity and enterprise search company Glean highlighted as part of its strategy. This move reflects OpenAI’s intent to broaden its offerings for enterprises and end users. The company has ambitious revenue targets, aiming to increase its earnings from $3.7 billion this year to $11.6 billion by 2025, signalling a strong push for growth in the competitive AI landscape.
While OpenAI’s request for exclusive commitments from investors is not legally binding, it underscores the company’s strategy to capitalise on its strong market position in a highly competitive environment where securing funding is crucial. Typically, venture capitalists steer clear of investing in direct competitors, but OpenAI’s approach is somewhat atypical. The situation is further complicated by late-stage investors like SoftBank and Fidelity, which have invested in both xAI and OpenAI, blurring the lines in the competitive landscape. This dynamic highlights the challenges and complexities investors face in navigating the rapidly evolving AI sector.
OpenAI’s request does not affect its past investors or their existing investments but could influence future fundraising efforts for both OpenAI and its listed competitors. The Financial Times and Wall Street Journal were among the first to report on the names of the companies involved.
Durk Kingma, one of the lesser-known co-founders of OpenAI, has announced he is joining AI research company Anthropic. Kingma, who will work remotely from the Netherlands, expressed excitement about aligning with Anthropic’s mission to develop AI systems responsibly. However, he did not specify his exact role within the company.
Kingma, with a PhD in machine learning from the University of Amsterdam, played a crucial role in developing generative AI models like DALL-E and ChatGPT during his time at OpenAI. After leaving OpenAI in 2018, he rejoined Google Brain before its merger with DeepMind in 2023, while also working as an angel investor for AI startups.
Kingma’s hiring is part of a broader trend, with several key figures from OpenAI moving to Anthropic in recent months, including safety lead Jan Leike and co-founder John Schulman. Anthropic, led by former OpenAI VP Dario Amodei, continues to attract top talent as it positions itself as a more safety-conscious alternative in the AI industry.
Apple has reportedly exited negotiations to participate in OpenAI’s upcoming funding round, expected to raise around $6.5 billion. The round is anticipated to close next week, according to the Wall Street Journal. Other major tech companies, including Microsoft and Nvidia, are still engaged in talks to invest.
Microsoft is expected to contribute approximately $1 billion to the round, after previously investing $13 billion into OpenAI. The funding round is part of the tech industry’s growing interest in generative AI, which has sparked an arms race among companies to secure a competitive edge.
OpenAI‘s fundraising is projected to value the company at over $100 billion, largely driven by the success of ChatGPT. The AI model’s launch in late 2022 encouraged companies across multiple sectors to invest heavily in the technology to remain competitive.
OpenAI declined to comment on the news, while Apple did not respond to requests for clarification regarding its withdrawal from the negotiations.
SoftBank’s Vision Fund plans to invest $500 million in OpenAI’s latest funding round, according to a report by The Information. While SoftBank declined to comment, OpenAI did not respond immediately. The AI company is seeking to raise $6.5 billion through convertible notes, as reported by Reuters earlier this month.
Before SoftBank’s investment, OpenAI was valued at $150 billion, although this valuation is subject to change based on the company’s ability to restructure its corporate structure and lift the profit cap for investors. This marks SoftBank’s first investment in the firm led by Sam Altman.
OpenAI is reportedly considering a shift to a for-profit corporation that would function independently of its non-profit board. This transition aims to provide more flexibility and access to capital as the company seeks to enhance its business model and attract further investments.
OpenAI’s board is considering compensating CEO Sam Altman with equity, though no decision has been made, according to board chair Bret Taylor. The idea emerged as part of a broader conversation on restructuring the company’s core business into a for-profit benefit corporation. Altman holds no stake in OpenAI despite co-founding the organisation in 2015. His initial decision to forgo equity was rooted in the need for a majority of disinterested directors on the board.
The restructuring plan suggests that Altman could finally receive equity in the new for-profit entity. However, this transition comes amid sudden changes at the company, including the departure of several key executives. Speaking at a conference in Italy, Altman dismissed any connection between these exits and the ongoing discussions about the corporate restructuring.
As OpenAI navigates this pivotal transformation, it is also amid negotiations to raise $6.5 billion, with support expected from major players such as Microsoft, Nvidia, and Apple. The company’s potential valuation of $150 billion hinges on successfully executing its restructuring, which may include lifting a cap on investor returns.
Why does it matter?
Once restructured, the for-profit entity would no longer be under the control of the non-profit board that has overseen OpenAI since its inception. However, the non-profit organisation will continue to exist, holding a minority stake in the newly formed for-profit company. The shift signals a significant evolution for the AI giant as it seeks to attract further investment and scale its ambitions.
OpenAI’s official press account on X was hacked by cryptocurrency scammers, promoting a fraudulent blockchain token, ‘$OPENAI.’ The scammers posted a message claiming the fake token would grant users access to future OpenAI beta programs. The post linked to a phishing website designed to steal cryptocurrency wallet credentials from unsuspecting users. Despite the scam being evident, the post and the associated site remained active, with comments disabled to make the hack less noticeable.
This incident is part of a larger pattern, with OpenAI leadership accounts also targeted in similar phishing campaigns earlier this year. In June 2023, OpenAI CTO Mira Murati’s account was hacked, posting a nearly identical message about the non-existent “$OPENAI” token. Other key OpenAI staff, such as chief scientist Jakub Pachocki and researcher Jason Wei, were also hacked recently, further exposing vulnerabilities.
Cryptocurrency scams targeting high-profile X accounts have become increasingly common. In previous years, accounts belonging to Apple, Elon Musk, and Joe Biden were compromised to promote scams. These fraudulent campaigns often use fake offers or phishing schemes to steal funds from victims by tricking them into sending cryptocurrency to scam wallets.
Cryptocurrency scams have cost United States citizens $5.6 billion in 2023 alone, a significant increase from the previous year. With over 50,000 cases reported in the first half of 2024, losses have already reached $2.5 billion, according to the Federal Trade Commission, marking an alarming rise in the threat posed by such scams.
OpenAI is expanding its Advanced Voice Mode (AVM) to more ChatGPT users, beginning with those subscribed to the Plus and Teams plans, while Enterprise and education customers will gain access next week. The updated AVM includes a redesigned interface, now featuring a blue animated sphere, and introduces five new voices: Arbor, Maple, Sol, Spruce, and Vale. These additions bring the total voice options to nine, replacing ‘Sky,’ which was removed after legal issues arose over its similarity to actress Scarlett Johansson’s voice.
The AVM update also includes improvements like better accent recognition and smoother conversations. OpenAI has incorporated customisation options, including Custom Instructions, which allow users to personalise ChatGPT’s responses, and Memory, which enables ChatGPT to recall past conversations. However, previously showcased features such as video and screen-sharing remain unavailable, with no confirmed timeline for their release.
Despite the updates, AVM is not yet available in certain regions, including the EU, the UK, and several others. OpenAI is actively refining the feature based on early user feedback, working to resolve glitches and improve overall performance for a smoother experience.
Generative AI is significantly more energy-intensive than traditional search engines, according to researcher Sasha Luccioni, who has raised concerns about the environmental impact of the technology. Generating new information requires vast computing power and energy, particularly for models like ChatGPT, which rely on extensive data training.
The AI and cryptocurrency sectors consumed nearly 460 terawatt hours of electricity in 2022, around two percent of global production, according to the International Energy Agency. Luccioni, a leading expert on AI’s climate impact, has developed tools to quantify the carbon footprint of AI technologies, helping developers make informed decisions.
Efforts to mitigate the environmental consequences of AI are underway. Luccioni is working on a certification system to rate the energy efficiency of AI models, aiming to encourage more sustainable practices. Transparency from tech giants like Google and OpenAI is essential, as their greenhouse gas emissions have surged due to AI development.
The solution, Luccioni argues, lies in a combination of government legislation, increased transparency, and better public understanding of AI’s limitations and environmental costs. She advocates for ‘energy sobriety’ by using AI tools more judiciously and making environmentally conscious decisions.
A group of technology experts has launched a global call for ‘Humanity’s Last Exam‘ aiming to push AI systems to their limits by posing the most difficult questions possible. The Center for AI Safety (CAIS) and Scale AI are leading an initiative to establish when AI achieves expert-level capabilities. Current benchmark tests have become too easy for many AI models, so this effort aims to create a new exam that emphasises abstract reasoning, an area in which AI still faces challenges. The organisers hope this new exam will remain relevant as AI technology evolves.
The demand for more rigorous tests comes after OpenAI released its newest model, OpenAI o1, which has shown strong performance in traditional reasoning benchmarks. Dan Hendricks, executive director of CAIS, stated that AI systems like Anthropic’s Claude model had significantly improved standard tests, rendering these benchmarks less valuable. However, AI has struggled with more intricate tasks like planning and visual pattern recognition, highlighting the necessity for more advanced assessments.
The exam will include over 1,000 crowd-sourced questions that are challenging even for non-experts. Its goal is to prevent AI from simply memorising answers by keeping some questions private. Participants have until 1 November to submit questions, and there will be rewards for the best contributions. While the exam is designed to test AI thoroughly, questions about weapons will be excluded to avoid potential risks.
