TOP AI incidents of 2024
2024 has been nothing short of a whirlwind year. Geopolitical conflicts showed no signs of resolution, and the U.S. presidential election added another layer of global attention and debate. Amid this backdrop, the race for artificial intelligence (AI) continued to escalate at an unprecedented pace. Despite calls earlier to pause AI development and assess its risks, big tech companies doubled down, funneling massive investments into research, development, and deployment.
It’s clear that the momentum behind AI is unstoppable, but here’s the reality and I repeat again this year: AI isn’t just a topic for data scientists, engineers, governments, or tech giants. It impacts everyone — whether you’re a consumer, policymaker, or simply someone navigating the modern digital world.
Yet, as we rush to embrace AI’s transformative potential, I fear we aren’t paying enough attention to its risks. The consequences of ignoring these could be significant, affecting everything from jobs and privacy to global security. That’s why I’ve written this article: to provide a comprehensive summary of AI’s key developments in 2024 and highlight the risks and opportunities that everyone needs to understand.
By reading this, you’ll gain insights into how AI evolved this year and why it’s essential for all of us — not just experts — to engage in the conversation about its future.
January
- Duolingo cuts workers as it relies more on AI (Topic: labour)
Duolingo, an AI company specializing in language learning, made significant changes to its workforce. The company decided to reduce approximately 10% of its contractor workforce. This move was driven by Duolingo’s adoption of AI models, including OpenAI’s GPT-4, for content production and translations. By leveraging AI technology, Duolingo aims to enhance its language learning platform and improve user experiences.
While AI adoption offers efficiency gains, it also poses risks to human jobs. Companies like Duolingo are streamlining operations by replacing human contractors with AI models. As AI becomes more prevalent, there’s concern about job displacement and the need for retraining workers to adapt to changing roles. Balancing technological progress with workforce stability remains a critical challenge.
Source: https://www.washingtonpost.com/technology/2024/01/10/duolingo-ai-layoffs/
- Taylor Swift AI images prompt US bill to tackle non-consensual, sexual deepfakes (Topic: privacy, safety & security)
Sexually explicit AI-generated deepfake images of American musician Taylor Swift circulated on social media platforms, including 4chan and X . These images prompted Microsoft to enhance its text-to-image model in Microsoft Designer to prevent future abuse. Several artificial images of Swift, some of a sexual or violent nature, quickly spread, with one post reportedly viewed over 47 million times before its eventual removal. Advocacy groups, US politicians, and Microsoft CEO Satya Nadella expressed concern, and it has been suggested that Swift’s influence could result in new legislation regarding the creation of deepfake pornography.
Deepfake technology can be misused to create harmful content, including non-consensual pornography. As AI algorithms become more sophisticated, the risk of malicious actors using them for harmful purposes increases. The Taylor Swift incident highlights broader risks associated with AI. In the context of safety and security, AI can be weaponized to create non-consensual deepfake material, leading to emotional distress, reputational damage, and legal implications. As AI technology evolves, safeguarding against misuse becomes crucial to maintaining a safe online environment.
February
- AI hiring tools may be filtering out the best job applicants (Topic:bias & fairness)
Companies increasingly rely on AI-driven hiring platforms to screen job candidates. These tools include body-language analysis, vocal assessments, gamified tests, and CV scanners. An IBM survey revealed that 42% of companies were already using AI screening to improve recruiting and human resources, while another 40% were considering integrating this technology. The hope was that AI recruiting tech would reduce biases in the hiring process. However, experts warn that these tools may inaccurately screen highly qualified applicants, potentially excluding the best candidates.
Despite the intention to eliminate bias, AI hiring tools sometimes exacerbate the problem. Some qualified candidates have been rejected due to flawed AI evaluations. For instance, a make-up artist in the UK lost her job after an AI tool scored her body language poorly, even though she performed well in skills evaluation. The lack of transparency in how candidates are evaluated by these tools adds to the concern. Additionally, many selection algorithms are trained on specific employee profiles, potentially filtering out candidates with diverse backgrounds or credentials.
Source: https://www.bbc.com/worklife/article/20240214-ai-recruiting-hiring-software-bias-discrimination
- Air Canada ordered to pay customer who was misled by airline’s chatbot (Topic: Accountability)
Air Canada’s chatbot provided incorrect information to passenger Jake Moffatt. The chatbot promised a discount that wasn’t actually available, assuring Moffatt that he could book a full-fare flight for his grandmother’s funeral and then apply for a bereavement fare afterward. However, when Moffatt applied for the discount, the airline denied it, claiming that the chatbot had been wrong — the request needed to be submitted before the flight. Air Canada argued that the chatbot was a “separate legal entity” responsible for its own actions. Nevertheless, the British Columbia Civil Resolution Tribunal ruled in favour of consumer, ordering Air Canada to pay him $812.02 in damages and tribunal fees. The decision emphasised that airlines are responsible for all information on their websites, whether from static pages or chatbots.
This case sets a precedent for companies relying on AI and chatbots for customer interactions. It establishes the principle that businesses cannot hide behind chatbots — they are liable for what their technology says and does. As more travel companies embrace AI, including chatbots, there are risks associated with inaccurate information and flawed decision-making. Airlines and other businesses must exercise caution when integrating AI tools to avoid legal and reputational consequences.
Source: https://www.theguardian.com/world/2024/feb/16/air-canada-chatbot-lawsuit
March
- Israel Implements Facial Recognition Program in Gaza Strip Amid Controversy (topic: privacy, bias & fairness)
Israel has implemented a large-scale facial recognition initiative in the Gaza Strip without the awareness or consent of Palestinians. Launched in response to the October 7th attacks, this program utilises technology from Google Photos and a specialised tool developed by Corsight. Its primary goal is to identify individuals associated with Hamas. However, the system has been marked by instances of mistaken identity, leading to wrongful detentions of civilians misidentified as Hamas militants.
The deployment of facial recognition technology in Gaza raises ethical concerns and accuracy issues. The software, initially used to search for Israelis taken by Hamas, has increasingly been employed to locate members of Hamas and other militant groups. The reliance on this technology can result in false positives, potentially violating human rights and causing harm to innocent individuals. Ensuring transparency, accountability, and safeguards against miss-identification is crucial when implementing AI-driven surveillance systems.
- NYC’s government chatbot is lying about city laws and regulations (topic: disinformation)
The New York City government’s “MyCity” chatbot, launched as a pilot program in October 2023, was designed to provide business owners with information from over 2,000 NYC Business web pages and articles. However, a recent report by The Markup and local nonprofit news site The City found that the chatbot is giving dangerously wrong information about basic city policies. For instance, it incorrectly stated that NYC buildings are not required to accept Section 8 vouchers, even though an NYC government info page clearly states that landlords must accept these housing subsidies without discrimination. The chatbot also provided incorrect information regarding worker pay, work hour regulations, and industry-specific details like funeral home pricing.
The MyCity chatbot’s inaccuracies highlight the risks associated with AI-driven systems. Powered by Microsoft Azure, the chatbot relies on statistical associations across millions of tokens to predict the next word in a sequence, lacking true understanding of the underlying information. This approach can lead to incorrect answers, especially when factual information isn’t precisely reflected in the training data. As AI systems become more integrated into critical services, ensuring accuracy and transparency is crucial to prevent misinformation and potential legal violations.
source: https://arstechnica.com/ai/2024/03/nycs-government-chatbot-is-lying-about-city-laws-and-regulations/
April
- Major U.S. newspapers sue Microsoft, OpenAI over alleged copyright violations (topic: copyright)
A group of eight major U.S. newspapers, including the New York Daily News, Chicago Tribune, and Denver Post, has filed a lawsuit against ChatGPT-maker OpenAI and Microsoft. The newspapers allege that these tech giants illegally used millions of their copyrighted articles to train sophisticated AI models, such as OpenAI’s ChatGPT and Microsoft’s Copilot. The lawsuit was filed in a New York federal court, and the newspapers are owned by investment company Alden Global Capital.
The lawsuit highlights a broader concern about the use of copyrighted content for training AI systems. While tech companies argue that this practice falls under the “fair use” doctrine of American copyright law, it raises questions about the balance between innovation and respecting intellectual property rights. As AI systems become more prevalent, ensuring fair compensation for content creators while advancing technology remains a challenge. The case underscores the need for clearer guidelines and ethical considerations in AI development.
- How AI is facing discrimination concerns (Topic: bias & fairness)
Facial recognition technology (FRT) has led to legal claims, including one of the first employment tribunal cases involving AI. An Uber Eats delivery driver was dismissed after FRT failed to identify him. The driver claimed that the technology is less accurate for non-white individuals, putting them at a disadvantage. A report to the UN highlighted evidence that automatic FRT algorithms disproportionately misidentify black people and women.
The accuracy of FRT systems poses a significant risk. Algorithms can exhibit racial or sexual discrimination, as seen in cases where they misidentify darker-skinned women more frequently. Employers using FRT must be transparent about its application to avoid discrimination claims. Understanding how AI works and addressing biases is crucial to prevent unintended consequences.
source: https://magazine.dailybusinessgroup.co.uk/2024/04/26/how-ai-is-facing-discrimination-concerns/
May
- Microsoft’s AI Push Jeopardizes Climate Goals as Emissions Surge (topic: sustainability)
Microsoft pledged in 2020 to remove more carbon dioxide from the atmosphere than it emits by 2030, with the goal of reversing its lifetime carbon emissions by 2050. However, the company’s carbon emissions increased by 30% in 2023 compared to 2020. The primary reason for this surge is the rapid construction of data centres, which require carbon-intensive building materials. These data centres are crucial for running and supporting AI models like large language models, such as OpenAI’s ChatGPT and Google’s Gemini, which are seeing widespread adoption worldwide.
Microsoft’s relentless push to be a global leader in AI is putting its climate goals in jeopardy. The company’s total planet-warming impact is now about 30% higher than it was in 2020. While AI services like ChatGPT and similar models are in high demand, they rely on power-hungry data centres built from carbon-intensive materials. Balancing AI advancements with sustainability remains a challenge for tech companies, including Microsoft, Google, Meta, and Amazon, all of which have experienced increased carbon emissions despite setting climate goals.
source: https://www.cnet.com/tech/mobile/microsofts-ai-push-jeopardizes-climate-goals-as-emissions-surge/
- Autonomous taxi company pays woman millions after she was dragged across San Francisco street (topic: safety & security)
Photo by Lexi Anderson on Unsplash
A woman in San Francisco was hit by a self-driving taxi operated by General Motors’ autonomous car company, Cruise. The incident occurred last year when the woman was struck by a human hit-and-run driver and propelled into the path of the Cruise robo-taxi. Despite the vehicle’s attempt to pull over, it continued for about 20 feet (around 6 meters), pinning the pedestrian to the bottom of the car. The woman sustained traumatic injuries and was hospitalised. Cruise has now settled the lawsuit with the woman, agreeing to pay her between $8 to $12 million.
This incident highlights the risks associated with autonomous vehicles. While self-driving technology aims to improve safety, accidents like this raise concerns about detection capabilities, emergency responses, and human-machine interaction. As companies continue to develop and deploy AI-driven systems, ensuring robust safety measures and addressing potential failures becomes crucial to prevent harm to pedestrians and passengers.
source: https://uk.news.yahoo.com/autonomous-taxi-company-pays-woman-182513755.html
June
- Meta’s privacy policy lets it use your posts to train its AI (topic: privacy)
Meta, the parent company of Facebook and Instagram, recently proposed a policy change that would allow it to use publicly posted content from users in the European Union (EU) and the United Kingdom (UK) to train its generative AI (genAI) models. This includes public posts, images, comments, and intellectual property. However, Meta clarified that it would not use private posts or messages for AI training. Users in the EU and UK have the option to opt out of having their content used for AI training by filling out an objection form.
While Meta’s approach aims to be transparent, concerns remain. By using publicly available information, including colloquial phrases and local references, Meta builds its foundational AI model. However, this raises privacy issues, especially when users may not be fully aware of how their data is being utilized. Regulators in the EU and UK have pushed back, emphasizing privacy concerns. Additionally, the potential for unintended biases and ethical implications in AI training remains a risk that companies like Meta must address.
- AI could strain Texas power grid this summer (topic: sustainability)
Texas faces power grid anxiety due to factors like extreme heat, aging power plants, and integrating renewable energy. Now, energy-hungry computer data centres pose an additional risk. Many of these centres mine cryptocurrency, but an increasing number are built to support AI systems. These centres are drawn to Texas due to low energy costs, minimal regulation, and a booming economy. Running an AI search consumes significantly more energy (10 to 30 times) than a traditional Google search. While state officials have worked to strengthen the power grid, data centres can be built in just months, presenting a challenge for grid operators.
The surge in AI data centres strains Texas’ already fragile energy system. These centres consume substantial energy, yet produce few jobs. State lawmakers are questioning whether data centres are worth the strain on the grid. Crypto-mines and data centres could ultimately lead to higher costs for Texans, as the state grapples with meeting growing energy demand. It’s clear that balancing the benefits of AI development with its impact on energy infrastructure remains a significant challenge.
July
- It’d be comedy if not real & tragic: Economic Survey on developed nations pressuring developing countries to reduce emissions (topic: sustainability)
The Economic Survey 2024 criticized developed nations for pressuring developing countries to reduce emissions while their own AI adoption drives a surge in energy demands. The survey highlights the irony of developed nations focusing on prospective emissions from developing countries while increasing their own energy consumption through AI. This obsession with AI risks exacerbating inequality and poverty in developing countries by prioritizing emissions over economic growth.
- Senior Republican Senator demands that OpenAI prove it doesn’t silence staff (topic: transparency)
A senior Republican senator has demanded that OpenAI provide documents proving it does not silence employees who wish to share concerns with federal regulators about the development of its AI tools. This request follows a Washington Post report and highlights growing bipartisan pressure on OpenAI to ensure the safe development of its AI technologies. The senator’s letter underscores the importance of transparency and accountability in the AI industry, particularly regarding employee rights and safety protocols.
The call for OpenAI to disclose its practices comes amid broader concerns about the ethical implications and potential risks associated with AI development. Lawmakers are increasingly scrutinizing AI companies to ensure they adhere to safety standards and do not suppress internal dissent. This move reflects a growing recognition of the need for robust oversight and regulation in the rapidly evolving field of artificial intelligence.
source: Senior Republican Senator demands that OpenAI prove it doesn’t silence staff — The Washington Post
August
- AI lending will make finance deals even more unfair for women — here’s how this can be avoided (topic: fairness, bias)
Research shows that women often receive less favorable loan terms than men, a trend that AI could worsen if not properly managed. A study of over 50,000 car loans in Canada revealed that the expected utility of loans was significantly lower for women than for men. AI, when used to optimize commissions for sales representatives, could further disadvantage women by assuming they are more tolerant of worse offers, leading to a greater disparity in loan terms.
Bias in AI creates significant risks by perpetuating and amplifying existing inequalities. If AI systems are trained on biased data, they can reinforce discriminatory practices, making it harder for marginalized groups to access fair financial services. This not only affects individual borrowers but also undermines trust in financial institutions and the broader AI technology. To mitigate these risks, it is crucial to design AI systems that prioritize fairness and transparency, ensuring that they do not disproportionately harm disadvantaged groups.
source: AI lending will make finance deals even more unfair for women — here’s how this can be avoided
- Women face greater risk of job displacement from automation (topic: fairness, bias)
Research by Code First Girls and Tech Talent Charter indicates that job automation is 40% more likely to affect women than men. This disparity is partly due to the fact that the majority of software engineers are men, which can lead to AI systems being developed with inherent biases. The report emphasizes the importance of ongoing training, such as upskilling and reskilling, to help women retain their jobs and support their career growth in tech. The pandemic has further exacerbated these biases, as women were more likely to be furloughed and take on caregiving responsibilities, making them more vulnerable to job displacement.
The risk behind this issue is the perpetuation of gender biases in AI development, which can lead to significant job losses for women. If AI systems are developed with inherent biases, they can reinforce existing inequalities, making it harder for women to secure and retain jobs in the tech industry. To mitigate these risks, it is crucial to implement inclusive training programs and ensure diversity in AI development teams.
source: Women face greater risk of job displacement from automation | Computer Weekly
September
- 387 apprehended for deepfake sex crimes this year, 80% teenagers (topic: bias, safety)
The National Police Agency received 812 reports of deepfake-related sex crimes from January to September in South Korea, with a significant increase in cases since a special crackdown began in late August. Among the apprehended suspects, a large majority were teenagers, including some under the age of 14 who are legally exempt from criminal punishment. The article highlights specific cases, such as two teenage men arrested for selling deepfake content on Telegram and a man in his 30s who manipulated images of colleagues and friends to create and distribute deepfake videos.
The rise in deepfake sex crimes, particularly among teenagers, underscores the growing severity of this issue in South Korea. The police’s intensified efforts to combat these crimes reflect the urgent need to address the misuse of deepfake technology.
- Meta CEO Mark Zuckerberg Dismisses Creators’ Rights; Tries To Justify Exploitative AI Practices (topic: intellectual property)
In a recent interview, Zuckerberg argued that content creators should allow AI companies to use their work for free, suggesting that creators overestimate the value of their content. This perspective has sparked significant backlash, especially amid ongoing lawsuits against AI companies for scraping copyrighted content without permission. Major record labels, authors, and other content creators have been pushing back against this practice, raising important legal and ethical questions about the use of their work in AI development.
Zuckerberg’s comments reflect a broader attitude among tech executives who see little value in compensating creators for the use of their work in AI training. This aligns with the positions of other tech leaders like OpenAI CEO Sam Altman and Microsoft AI CEO Mustafa Suleyman, who argue that publicly available content should be treated as fair use. However, many copyright holders disagree, leading to lawsuits aimed at clarifying how their work is being used without permission. The debate highlights the tension between technological advancement and the rights of content creators in the rapidly evolving AI landscape.
October
- Teenager took his own life after falling in love with AI chatbot. Now his devastated mom is suing the creators (topic: safety)
A tragic incident happened where a teenager named Sewell Setzer III took his own life after developing an emotional dependency on an AI chatbot named Daenerys, themed after a character from Game of Thrones. Sewell’s mother, Megan Garcia, has filed a lawsuit against Character Technologies, the creators of the chatbot, accusing them of negligence, intentional infliction of emotional distress, and wrongful death. The lawsuit claims that Sewell’s interactions with the chatbot, which included sexual content despite him identifying as a minor, led to his deteriorating mental health and eventual suicide. The company has expressed condolences and mentioned implementing new safety measures, but the lawsuit seeks to hold them accountable for the harm caused.
The case highlights the potential dangers of AI chatbots, especially for vulnerable individuals like teenagers. It underscores the need for stringent safety measures and oversight to prevent such tragedies.
- Instagram’s new AI-driven feed is causing people to spend way more time on their phones, Meta reveals (topic: Transparency, safety)
According to Meta’s latest quarterly results, the AI improvements have resulted in an 8% increase in time spent on Facebook and a 6% increase on Instagram. This translates to an additional 10 to 15 hours per year for the average user. Meta CEO Mark Zuckerberg announced plans to introduce a new feed that will feature exclusively AI-generated or AI-summarized content, which he believes will be an exciting addition to the platforms.
Despite the increase in user engagement, Meta’s share price fell in after-hours trading. Zuckerberg expressed confidence that AI-driven content will be a significant trend in the coming years, contributing to the company’s growth. However, the reliance on AI to boost user engagement raises questions about the potential impact on user behavior and the ethical implications of AI-generated content.
November
- Microsoft’s AI Recall Tool Faces Another Delay Amid Privacy Concerns (topic: privacy)
Initially introduced in May for Windows 11, Recall is designed as a “time machine” that captures and organizes screenshots of everything displayed on a user’s screen, making it searchable. However, privacy advocates raised concerns about the tool, leading Microsoft to postpone its rollout for further review. The company now plans to release Recall to Windows Insiders on Copilot-enabled PCs by December, pending additional internal evaluations.
Microsoft’s delay highlights the ongoing challenges tech companies face in balancing innovative AI features with user privacy and security. Despite the potential benefits of Recall, the need for stringent privacy measures has become paramount. Microsoft has emphasized that the tool will be opt-in only, with sensitive data always encrypted and protected. This cautious approach reflects the broader industry trend of navigating the complexities of generative AI while ensuring user trust and data protection.
- Will LinkedIn’s AI HR assistant select the right candidates? (topic: privacy, fairness, transparency)
The “Hiring Assistant” from LinkedIn automates repetitive steps like candidate pre-selection, using a competency-based model that prioritizes skills over diplomas. This AI tool, currently in the testing phase with companies like AMD, Canva, Siemens, and Zurich Insurance, is expected to provide automated follow-up with candidates by 2025, enhancing the efficiency of the recruitment process.
However there are potential risks such as the creation of fake recruiter profiles by cybercriminals to extract personal information from job seekers. This underscores the need for robust security measures to protect users. While the AI assistant has the potential to revolutionize recruitment, it is crucial to address these security concerns to ensure the bias, transparency, safety and privacy of LinkedIn’s over one billion members worldwide.
source: Will LinkedIn’s AI HR assistant select the right candidates? | The Star
December
- AI Generated Police Reports Raise Concerns Around Transparency, Bias (topic: bias, fairness, transparency)
There are growing use of AI-generated police reports by some police departments, raising significant concerns about transparency, bias, and the reliability of these reports. The ACLU highlights four main issues: the inherent quirks and biases of AI, the potential for AI to overwrite important details from officers' memories, the lack of transparency in how these AI systems operate, and the loss of accountability when officers rely on AI to draft reports. The ACLU argues that AI-generated reports could lead to inaccuracies and biases being introduced into the criminal justice system, potentially affecting the outcomes of criminal investigations and prosecutions.
The ACLU's white paper emphasizes that police reports are crucial in determining innocence, guilt, and punishment, and that introducing AI into this process could undermine the integrity of the justice system. This underscores the need for transparency and accountability in the use of AI in law enforcement to ensure that civil liberties and rights are protected.
Source: AI Generated Police Reports Raise Concerns Around Transparency, Bias | ACLU
- The Terminator’s Vision of AI Warfare Is Now Reality (topic: human right, safety)
There is an alarming reality of AI-powered warfare. Autonomous weapons and AI-driven drones are being deployed in real-world conflicts, such as in Gaza and Ukraine. These technologies, initially seen as science fiction, have become a twenty-first-century reality, raising significant ethical and humanitarian concerns. The use of AI in warfare has led to devastating consequences, particularly in Gaza, where Israeli military operations have employed AI-powered drones to target and kill civilians.
This underscores the psychological trauma inflicted on populations living under constant drone surveillance and the ethical implications of using AI to make life-and-death decisions. It calls for greater scrutiny and regulation of AI technologies in military applications to prevent further human rights abuses and ensure accountability.
source: The Terminator’s Vision of AI Warfare Is Now Reality
Conclusion
In conclusion, 2024 has been a pivotal year for AI, marked by significant advancements and equally substantial challenges. From job displacement and privacy concerns to ethical dilemmas and the potential for misuse, the incidents throughout the year underscore the urgent need for robust regulations and ethical guidelines.
As AI continues to integrate into various aspects of our lives, it is crucial to balance innovation with responsibility, ensuring that the technology benefits society while safeguarding against its risks.
**Disclaimer: The views expressed in this article are solely my own and do not reflect the opinions, beliefs, or positions of my employer. Any opinions or information provided in this article are based on my personal experiences and perspectives. Readers are encouraged to form their own opinions and seek additional information as needed.**
