Tech On Me: Will Singapore succeed in banning deepfakes?

This week's focus 

Singapore is exploring ways to regulate deepfake content and potentially introducing a temporary ban due to concerns that AI may significantly blur the lines between fact and fiction ahead of elections. 

Minister for digital development and information Josephine Teo mentioned at the Reuters Next Apac conference that various countries, like South Korea, are implementing measures such as a 90-day ban on political AI content before elections. 

However, Teo noted that Singapore's short election period requires different solutions. The country addresses AI-generated misinformation with laws like the Protection from Online Falsehoods and Manipulation Act (Pofma), which can be applied to AI-generated fake news. 

Teo emphasised the need for regulations to close loopholes in the law regarding AI-created falsehoods. In addition, she discussed Singapore's ambition to become a global AI player, highlighting the importance of talent, data access frameworks, and expanding data centre capacity to support AI development.

My take: As AI technology advances, the potential for misuse, especially in the political arena, is significant. For example, AI-generated videos featuring Singapore’s former prime minister, Lee Hsien Loong, discussing international issues and foreign leaders have emerged, and, more sinisterly, British female politicians have fallen victim to AI-generated fake pornography, with their faces used in explicit images. 

These images, some of which have been online for years and garnered hundreds of thousands of views, range from crude Photoshops to more sophisticated deepfakes created with AI technology.

On the other hand, during the recent Indian elections, politicians had no qualms about using deepfake audio and video of themselves to connect with voters, who were often unaware that they were interacting with a digital clone rather than the actual person. According to Wired, people living in rural areas frequently experience a heightened sense of importance when they receive personalised AI calls from individuals in high positions.

Regardless of how politicians use AI, I would like to see the law agile enough to address the unique characteristics of AI-generated deepfakes, which can be more sophisticated and more challenging to detect than traditional misinformation. However, I acknowledge that balancing the need for stringent regulations with the flexibility to adapt to new technological developments can be difficult.

In other news

Google pulls out of buying HubSpot

According to a Reuters source, Alphabet, the parent of Google, recently decided against pursuing a takeover of online marketing software company HubSpot. The discussions between Alphabet and HubSpot did not progress to due diligence and ended shortly after initial talks.

Reuters had earlier reported Alphabet's interest in acquiring HubSpot, valued at $25 billion. Such a deal would have been Alphabet's largest ever and might have attracted antitrust scrutiny from US regulators, who are increasingly wary of large tech companies expanding through acquisitions.

As CEO Yamini Rangan mentioned in May, HubSpot, known for its inbound marketing software designed for small and medium-sized businesses, has seen a decline in customer demand due to economic concerns over high interest rates. Bloomberg also reported that Google is no longer in acquisition talks with HubSpot.

My take: Alphabet's decision not to pursue the acquisition of HubSpot is a calculated move, as it seeks not to have more scrutiny on its businesses. US regulators have shown increasing resistance to allowing big technology firms to expand further through acquisitions due to concerns about monopolistic behaviour and the concentration of market power.

For example, the US Department of Justice is currently bringing a lawsuit against Google, accusing the platform of dominating the adtech world. The DOJ is also bringing a separate lawsuit against Google for monopolising search engines and advertising markets.

Confusion over video games and gambling in India

Indian video game companies are calling on the government to clearly differentiate video games from real money games to avoid regulatory confusion. This confusion has resulted in tax raids and investor complications, hindering the industry's growth. The firms are pushing for distinct definitions, tax reductions, and regulatory support to boost the sector and attract investment. 

They are asking the Indian government to replace the umbrella term ‘online games’ with distinct categories like ‘video games’ and ‘real money games’ after recent tax amendments placed real money games in a 28% tax bracket, while video games remained at 18%. 

The confusion has even led to police raids on video game parlours mistaken for gambling establishments and complicated dealings with international investors.

My take: This problem is not new and not just concentrated in India. I previously wrote in 2022 about how loot boxes in gaming involve new layers of regulatory complexity, as parallels with traditional gambling are often vague and difficult to interpret. Plus, existing gambling bodies are currently ill-equipped to deal with such challenges.

For example, Roblox, the widely popular online gaming platform, had to deal with a substantial class action lawsuit accusing the company of enabling an illicit gambling network that targets minors. The lawsuit alleges that Roblox's digital currency, known as ‘Robux’, has been misused for gambling purposes. Users under 18 can utilise Robux, which they buy with their parents' credit card, with each unit having an actual monetary value.  

I would like the government to create distinct categories to regulate both sectors without unnecessary overlap more effectively. This would prevent video game companies from facing undue scrutiny intended for gambling-related activities.

Should Meta compensate for news?

Meta has asserted that news content is not essential for combating misinformation on Facebook and Instagram, resisting pressures to compensate Australian media companies for news. Meta decided not to renew agreements to pay for news in March, following the expiry of contracts made under former prime minister Scott Morrison's government’s news media bargaining code.

Assistant Treasurer Stephen Jones is deliberating whether to enforce the bargaining code to compel Meta to negotiate payments with news providers or face fines of up to 10% of its Australian revenue. The Treasury is also considering other measures, such as mandating news content or leveraging the tax system, amid concerns that Meta might block news in Australia, as it has done in Canada since last August.

Meta claims there was no significant drop in user engagement in Canada after the news ban, suggesting that news content is replaceable. During a committee hearing, Meta’s director of policy in Australia, Mia Garlick, did not rule out blocking news if Meta was designated under the code.

My take: Despite what Meta wants the public to believe, the ongoing news ban in Canada has led to significant consequences, particularly for regional and local news sites that rely heavily on Facebook for audience traffic. Removing direct links to news articles has drastically reduced user visits to these sites, exacerbating concerns about the emergence of ‘news deserts’ in rural areas.

Although news outlets have attempted to circumvent the ban by posting article content without links or sharing screenshots, these efforts cannot fully recover the lost audience engagement or revenue generated from website traffic. The gap left by news content has often been filled by political discussions devoid of direct news references, leading to misinformation.

As someone working in news media, I am all for forcing Meta to pay news publishers, but I am wary that it might backfire. A more effective approach to supporting local news might involve using corporate tax revenues from digital platforms to fund public-interest journalism, ensuring a sustainable future for quality Australian news content.

China’s generative AI adoption surges

A recent survey from SAS and Coleman Parkes Research highlights that China is emerging as a global leader in adopting generative AI.

The survey, which included 1,600 industry decision-makers from the banking, healthcare, and telecommunications sectors, revealed that 83% of Chinese respondents use generative AI. This percentage surpasses the global average of 54% and the US rate of 65%.

China's progress in generative AI has accelerated since the release of OpenAI's ChatGPT in late 2022. The country's technological advancements are further evidenced by a report from the United Nations' World Intellectual Property Organisation, which showed China leading in GenAI patent filings, with over 38,000 patents between 2014 and 2023, compared to 6,276 by the United States.

Despite restrictions on many international generative AI services in China, the country has cultivated a robust domestic industry, featuring contributions from major tech companies like ByteDance and startups like Zhipu. The adoption of generative AI in Chinese enterprises is expected to grow, driven by competitive pricing that reduces the cost of large language model services.

However, the Shanghai Declaration on Global AI Governance, presented at the recent World Artificial Intelligence Conference, emphasises the need for transparent, traceable AI technologies. The declaration, issued by Shanghai's party secretary Chen Jining, calls for compliance with local laws and encourages international cooperation and openness in AI development.

My take: In the West, startups like OpenAI and Anthropic quickly evolved into significant players, while traditional tech giants like Microsoft, Amazon, and Google engaged primarily through partnerships and acquisitions. For example, Google's AI efforts stem from acquiring DeepMind, and Microsoft integrates OpenAI's innovations into its products. Similarly, Amazon collaborates with Anthropic.

Conversely, in China, major tech companies such as Baidu, Alibaba, and Tencent lead the generative AI efforts due to their substantial resources. These companies, collectively known as BAT, had the financial and technological capabilities to drive AI development, unlike startups which struggled to secure funding post-pandemic. 

However, Chinese tech giants didn't work in isolation; they partnered with universities and startups for specialised industry solutions. Alibaba, for instance, has invested in an AI startup called Baichuan, which is also backed by Tencent and Xiaomi. Similarly, Tsinghua University’s Zhipu and Kai-Fu Lee's 01.AI received funding from Alibaba and Tencent, highlighting significant collaborative efforts.

I believe generative AI will shift China from its traditional production-focused economy to a service-oriented industry. This shift will position China as a global leader in the service sector, with aspirations to increase the service industry's contribution to GDP, thereby boosting innovation, productivity, and living standards.

Kuaishou’s new short video generation feature 

Chinese short-form video platform Kuaishou recently introduced new features for its text-to-video model Kling AI at the World Artificial Intelligence Conference (WAIC) in Shanghai, enabling the creation of 10-second videos. Visitors at the event could test the tool by generating videos from simple prompts, showcasing Kling AI’s high accuracy. This has led to a surge of AI-generated videos on Chinese social media, including viral clips featuring popular characters performing modern tasks.

My take: Unsurprisingly, Kuaishou is ramping up its generative AI efforts as its main rival, ByteDance's Douyin and TikTok, has been on an AI hiring spree and launching new tools. In March 2024, Bytedance had over 320 generative AI-related job openings, including 130 positions focused on developing large language models (LLMs).

However, both Bytedance and Kuaishou have experienced a significant loss of top AI talent due to China's booming unicorn scene. This scene has produced four prominent "AI tigers," each securing substantial investments. These new unicorns, valued at over $1 billion, include Baichuan, Zhipu AI, Moonshot AI, and MiniMax. They all aim to emulate the success of OpenAI.

Looking ahead

South Korea, the next AI frontier? 

Hedge funds are increasingly investing in South Korean semiconductor firms like SK Hynix and Samsung Electronics, driven by the surge in demand for high-end memory chips fueled by the artificial intelligence (AI) boom. 

This shift comes as funds like Britain's Man Group, Singapore's FengHe Fund Management, and Hong Kong's CloudAlpha Capital Management and East Eagle Asset Management seek to capitalise on the undervalued stocks in South Korea compared to U.S. and Taiwanese counterparts.

The focus on South Korean firms is due to their leading role in producing high-bandwidth memory (HBM) chips, which are crucial for AI technologies. Hynix, a major supplier to Nvidia, benefits significantly from this demand, trading at a lower valuation than Taiwan's TSMC. 

South Korea's government support, including a $19 billion package and initiatives to boost shareholder returns, further bolsters these investments.

My take: Semiconductors are vital to South Korea’s export-driven economy. The sector's performance directly impacts the country's overall economic health. By investing heavily in AI and semiconductor technology, South Korea aims to maintain and potentially expand its market share in a highly competitive global landscape.

However, the success of this initiative will depend on effective implementation, collaboration with industry stakeholders, and the ability to navigate the complexities of global market dynamics.