A hacker infiltrated OpenAI’s internal messaging systems last year, stealing details about the design of its AI technologies, according to Reuters’ sources familiar with the matter. The breach involved discussions on an online forum where employees exchanged information about the latest AI developments. Crucially, the hacker needed access to the systems where OpenAI builds and houses its AI.
OpenAI, backed by Microsoft, did not publicly disclose the breach, as no customer or partner information was compromised. Executives briefed employees and the board but did not involve federal law enforcement, believing the hacker had no ties to foreign governments.
In a separate incident, OpenAI reported disrupting five covert operations that aimed to misuse its AI models for deceptive activities online. The issue raised safety concerns and prompted discussions about safeguarding advanced AI technology. The Biden administration plans to implement measures to protect US AI advancements from foreign adversaries. At the same time, 16 AI companies have pledged to develop the technology responsibly amid rapid innovation and emerging risks.
OpenAI’s ChatGPT macOS app was found to be storing user chats in plain text until recently, raising security concerns. The Verge reported that the AI firm has now released an update to encrypt conversations on macOS. The discovery was made by software developer Pedro Vieito, who noted that OpenAI was distributing the app exclusively through their website and bypassing Apple’s sandbox protections.
Sandboxing, which isolates an app and its data from the rest of the system, is optional on macOS, but is commonly used by chat applications to protect sensitive information. By not adhering to this security measure, the ChatGPT app exposed user chats to potential threats. Vieito highlighted the vulnerability on social media, showing how easily another app could access the unprotected data.
OpenAI acknowledged the issue and emphasised that users could opt out of having their chats used to train the AI models. The ChatGPT app, which was made available to macOS users on June 25, now includes encryption to enhance user privacy and security.
China’s leading internet search company, Baidu, has introduced the latest version of its large language model, Ernie 4.0 Turbo, to enterprise clients, Baidu Vice President Xie Guangjun announced at an event in Shanghai. The new model is priced at 30 yuan ($4.13) per million input tokens and 60 yuan per million output tokens. Baidu also announced significant price cuts of up to 83% for earlier Ernie models to stay competitive.
The adoption of Ernie has seen a substantial increase, with average daily usage rising by 150% in the second quarter. The surge in usage reflects the growing interest and reliance on advanced AI models in China.
The competitive landscape intensifies as significant tech firms and LLM startups in the country offer free credits or tremendous discounts to attract developers to their platforms. These price differences highlight the aggressive pricing strategies Chinese companies employ to capture market share in the rapidly evolving AI industry.
CompetitionAI, launched in November 2023 by Daniel Schwarz, a competition lawyer at Clifford Chance, alongside AI engineers Jonathan Halpern and Andrew Davidson, aims to revolutionise legal research for competition law practitioners. Dubbed ‘ChatGPT for competition lawyers’, the platform employs AI to swiftly navigate and extract relevant insights from vast volumes of competition authority guidance. Model like this one allows lawyers to obtain precise answers efficiently, verifying them against the original sources provided.
Since its inception, CompetitionAI has garnered over 600 registrations from users spanning more than 100 law firms and 60 corporates worldwide. Initially covering jurisdictions like the UK, EU, USA, and Australia, the platform has expanded its coverage to include Canada, Ireland, India, South Africa, Turkey, and Nigeria. The growth underscores its commitment to providing comprehensive competition law guidance on a global scale.
Daniel Schwarz, reflecting on the platform’s genesis, noted his transition from a decade-long career in law to a deep-seated fascination with AI’s potential to enhance legal practice. Partnering with AI experts, CompetitionAI utilises advanced models like GPT-4 to ensure precise and reliable information retrieval from regulatory texts and case law. The team continues to refine its AI capabilities, aiming to tailor the platform specifically to the evolving needs of competition lawyers.
Looking ahead, CompetitionAI plans to expand its team and capabilities through ongoing fundraising efforts, driven by escalating demand. Schwarz emphasises their dedication to leveraging AI to meet the nuanced demands of competition law, indicating a strategic focus on enhancing and scaling their platform to serve a growing international clientele effectively.
Japanese light rail and bus operator Yamaman Co has introduced facial recognition technology to its Jorudan Style Point&Pass ticketing system on the Yukarigaoka Line and local bus services. Passengers can now use the Eucalyptus Pass system by registering online with a photo and credit card details. At the stations, facial recognition cameras identify users, open barriers, and automatically charge their credit cards for the flat fare of ¥200 or a day ticket for ¥500.
Previously, passengers used magnetic tickets, but these machines are being updated to issue paper tickets with QR codes for occasional and non-registered travellers. The new technology builds on a successful 2021 pilot scheme on bus services, and suppliers J MaaS and Panasonic Connect aim to expand the system across Japan.
The implementation, costing around ¥60 million, was partially funded by a government subsidy and is expected to reduce ticketing costs by 30%. The koala theme of the transport services reflects the local presence of eucalyptus trees.
Chinese Premier Li Qiang has called for global cooperation and a more open mindset in the field of AI amid growing competition between China and the United States. Speaking at the World AI Conference (WAIC) in Shanghai, Li emphasised the importance of international collaboration, noting that each country has unique strengths in AI technology, data, and markets.
China has made significant strides in generative AI, with the UN reporting that Beijing has led the filing of generative AI patents globally. The progress has sparked concern in the US and other nations, who worry about potential espionage uses of Chinese technology. In April, the US revoked licenses for some chip exports after unveiling a Huawei computer powered by Intel AI technology.
Premier Li urged the promotion of cross-border data movement, free trade of equipment, and infrastructure connectivity to foster AI development. He also highlighted the need for international cooperation to address the risks associated with rapidly advancing AI technology, advocating for the establishment of global AI regulations and technical standards.
Google DeepMind and Harvard University researchers have developed a realistic virtual rat to study the neural circuits that control movement. The virtual rat’s brain, made up of artificial neural networks, was trained using hours of neural recordings from real rats.
This digital brain could predict and replicate the behaviour of actual rats, such as running or rearing up. The study identified key brain regions involved in movement and demonstrated that AI can simulate neural signals more accurately than older models.
Bridging the gap between AI and neuroscience, the project offers new ways to study brain functions and movements. However, this method allows researchers to tweak neural connections in the virtual rat to observe how changes affect behaviour, providing insights that are challenging to obtain through traditional lab experiments. By understanding how the brain commands muscle movements, the research could lead to advancements in both robotics and neuroscience.
Offering a platform to test hypotheses about brain function and behaviour quickly and efficiently, the virtual rat enables researchers to explore more complex tasks. The team plans to use these virtual rats to understand further how real brains generate intricate behaviours. Combining AI with biological data, the collaboration highlights the potential to uncover the mechanisms of brain function and movement.
A recent report indicates that China’s major venture capital investments increasingly focus on technology sectors that are aligned with government policies, such as AI and semiconductors. Despite an overall decline in funding, China accounted for 90% of global venture capital (VC) in the chip sector last year, investing $22.2 billion, more than double the $9.5 billion from 2022.
According to Preqin, a private investment data company, three significant deals, primarily involving partnerships between municipal authorities and Sino IC Capital, accounted for half of the semiconductor venture investments. Sino IC Capital manages the state-backed China Integrated Circuit Industry Investment Fund, also known as the Big Fund.
Overall, the report highlights three key takeaways. Firstly, China’s clean technology VC deals have surpassed those in the US in both value and volume for 2022 and 2023, especially in electric vehicles. Secondly, overall venture capital in Greater China has dropped sharply, a 42% decrease from the previous quarter, due to geopolitical tensions and fewer exit opportunities. Thirdly, foreign VC investments in Chinese firms have plummeted from $67 billion in 2021 to $19 billion in 2023, while domestic investments have remained stable.
Why does it matter?
The trend is clear evidence of the ongoing US-China tech supremacy war and the effect of US sanctions. With the US trade tariffs and restrictions on American venture capital firms prompting foreign investors to withdraw from the region, this report reveals how the Chinese government, in response, has stepped up and is backing state-funded entities to fill the void, with significant investments also coming from domestic tech companies.
WhatsApp is developing a new AI feature to create user avatars, following in the footsteps of Meta AI. According to WABetaInfo, this ‘Imagine Me’ feature will allow users to generate AI-based avatars by typing prompts in their chats. The feature was discovered in the latest WhatsApp beta for Android 2.24.14.13, available through the Google Play Beta programme.
Users can generate avatars by typing ‘Imagine me’ in the Meta AI chat or ‘@Meta AI imagine me’ in other chats. A screenshot from WABetaInfo shows how this feature might look. Once enabled, users must take setup photos, and the AI will create images based on the provided prompts. The resulting images are automatically shared in the conversation, with user privacy preserved.
The feature is optional and requires users to opt in through their settings. While currently available only in limited countries, it is still under development and cannot yet be tested by all users. WhatsApp aims to make Meta AI a more integral part of daily user interactions with this innovative avatar creation tool.
For the past two years, many unprofitable tech startups have faced significant challenges, leading to cost-cutting, mergers, or closures. However, startups focused on AI have been an exception, thriving amidst this broader downturn. The AI boom, which began in late 2022, has now become a strong counterpoint to the struggling startup landscape.
In the second quarter of this year, investors poured $27.1 billion into AI startups in the United States, nearly half of all US startup funding during that period. This surge in investment has driven overall startup funding to $56 billion, marking a 57% increase from the previous year and the highest three-month total in two years. AI companies are attracting significant funding, reminiscent of the tech investment frenzy of 2021.
Large financing rounds have bolstered the AI sector’s overall deal-making activity. Notable examples include CoreWeave raising $1.1 billion and later $7.5 billion in debt, Scale AI securing $1 billion, and xAI, founded by Elon Musk, obtaining $6 billion. Furthermore, this influx of capital has prompted venture capital investors to shift their focus, encouraging startups to aggressively pursue growth, particularly in AI.
Despite the excitement, building AI technology is expensive, requiring substantial investment in powerful computer chips and cloud storage. A recent analysis showed that AI startups spend an average of 22% of their computing costs, more than double that of non-AI software companies. As AI’s potential continues to generate significant hype, competition from big tech firms like Microsoft and Amazon may impact the ability of AI startups to secure such large sums in the future.
