Ethics & Policy
Shaping trustworthy AI: Early insights from the Hiroshima AI Process Reporting Framework

As artificial intelligence (AI) systems become increasingly integrated into our economies and societies, questions about their trustworthiness, safety, and societal impact have never been more pressing. To address these concerns, the Hiroshima AI Process (HAIP) was launched by G7 countries in 2023, culminating in a landmark international code of conduct for organisations developing advanced AI systems.
To support the implementation of this code, the OECD worked with experts from government, business, academia, and civil society to design a voluntary reporting framework. This framework aims to promote transparency, foster peer learning, and align AI development with democratic values and shared global interests. It was piloted in mid-2024 and officially launched in February 2025 at the French AI Action Summit.
In this blog post, we share preliminary insights from the first wave of organisational reports submitted through the HAIP framework. These findings offer a glimpse into how leading companies and institutions are addressing AI-related risks, promoting transparency, and aligning their systems with the public interest.
Participants from around the world, from big tech to small research institutes
Between February and April 2025, 19 organisations voluntarily submitted reports using the HAIP framework. Most were developers or deployers of advanced AI systems, with representation from Japan, the United States, Germany, Canada, Korea, Romania, and Israel. Participants included both large tech firms and smaller advisory or research institutions, providing a wide array of perspectives on how to responsibly develop and deploy AI.
What does the framework cover?
The HAIP reporting framework is organised around seven thematic areas that correspond to the actions of the G7 code of conduct:
- Risk identification and evaluation
- Risk management and information security
- Transparency reporting
- Organisational governance and incident management
- Content authentication and provenance
- Research and investment in AI safety
- Advancing human and global interests
Each section offers a structured approach for organisations to reflect on and communicate their practices, assisting in benchmarking against emerging international standards.
Key findings
1. Diverse approaches to AI risk evaluation
Organisations are using a variety of methods to identify and evaluate risks. These range from technical assessments and adversarial testing to stakeholder engagement and expert review. Larger companies often focus on systemic risks, such as societal bias or large-scale misuse, while smaller organisations emphasise sector-specific concerns.
Notably, the use of AI tools to test other AI systems is increasing. Several participants are also implementing structured risk frameworks and testing procedures, such as red and purple teaming, to identify vulnerabilities.
2. Layered risk management and security
All participants reported using multi-layered strategies to manage AI risks, including technical safeguards (such as model fine-tuning and secure testing environments), procedural controls, and real-time monitoring. Secure testing environments and robust cybersecurity practices—often aligned with ISO or NIST standards—are standard.
There is also attention to privacy and intellectual property: some organisations allow web content providers to opt out of data scraping, while others use privacy filters and contractual safeguards.
3. Transparency practices vary by sector
Consumer-facing companies frequently publish comprehensive reports, such as model cards and transparency disclosures. Business-to-business (B2B) firms often share such information privately with their clients. Overall, transparency regarding AI system limitations, risks, and updates is increasing.
However, practices regarding training data transparency remain inconsistent. While some open-weight models are thoroughly documented, others are less forthcoming, especially among B2B providers.
4. Stronger organisational governance
Most organisations are incorporating AI risk into their overall risk management systems or establishing dedicated AI governance frameworks. These initiatives generally encompass staff training, incident response protocols, and, in some instances, oversight at the board level.
Some organisations also depend on third-party audits or expert panels to validate their risk management practices. Incident management is becoming increasingly transparent, with several companies publishing updates and engaging in threat information-sharing initiatives.
5. Early steps on content authentication
Many organisations are now notifying users when they interact with AI, using disclaimers or interface design. However, the technical tools for verifying AI-generated content—such as watermarking or cryptographic credentials—remain in the early stages. Currently, adoption is led by a few major tech companies.
Despite limited uptake, several organisations are investing in these tools and contributing to international standard-setting efforts, including initiatives such as C2PA and NIST’s Synthetic Content Task Force.
6. Investing in AI safety and research
Organisations are increasingly investing in R&D to enhance AI safety, fairness, and interpretability. These efforts include internal safety labs, open-source tools, and collaborations with academia and civil society.
Several participants are contributing to the development of global safety norms through initiatives such as the Frontier Model Forum, the AI Safety Consortium, and the OECD’s Catalogue of Tools and Metrics for Trustworthy AI.
7. AI for public good and global priorities
Many organisations reported applying AI to address societal challenges, ranging from climate resilience and public health to education and accessibility. Examples include AI-powered diagnostics for tuberculosis, digital literacy programmes, and sustainable data infrastructure powered by renewable energy.
These initiatives are often linked to ESG priorities and the UN Sustainable Development Goals. Importantly, some organisations are tailoring tools for underserved regions and collaborating with civil society groups to design inclusive AI systems.
Ideas to improve the framework’s value for future reporting
Participants found the HAIP reporting process valuable for internal coordination, benchmarking, and clarifying roles and responsibilities around AI governance. While the framework has already proven useful, there is room to strengthen its relevance and usability:
- Keep pace with technological change: As AI capabilities evolve, the framework should adapt. Participants suggested updating it annually to reflect new risks and practices.
- Make reporting easier: Providing structured response options, drop-down menus, and clearer guidance could improve consistency and reduce barriers to participation.
- Offer tailored support: Role-specific modules for developers, deployers, and policymakers could help broaden engagement across the AI value chain.
- Encourage peer learning: Facilitated discussions and best practice exchanges among participants could strengthen the community of practice around trustworthy AI.
A foundation for global AI transparency
The HAIP reporting framework is a promising step towards greater transparency and accountability in the development of advanced AI systems. By voluntarily disclosing their practices, organisations not only build public trust but also help shape a shared understanding of what responsible AI looks like in practice.
The OECD will continue working with partners to refine the framework, support participating organisations, and expand its reach. These early insights mark the beginning of a broader journey towards operationalising the Hiroshima AI Process and advancing trustworthy AI on a global scale.
The post Shaping trustworthy AI: Early insights from the Hiroshima AI Process Reporting Framework appeared first on OECD.AI.
Ethics & Policy
5 interesting stats to start your week

Third of UK marketers have ‘dramatically’ changed AI approach since AI Act
More than a third (37%) of UK marketers say they have ‘dramatically’ changed their approach to AI, since the introduction of the European Union’s AI Act a year ago, according to research by SAP Emarsys.
Additionally, nearly half (44%) of UK marketers say their approach to AI is more ethical than it was this time last year, while 46% report a better understanding of AI ethics, and 48% claim full compliance with the AI Act, which is designed to ensure safe and transparent AI.
The act sets out a phased approach to regulating the technology, classifying models into risk categories and setting up legal, technological, and governance frameworks which will come into place over the next two years.
However, some marketers are sceptical about the legislation, with 28% raising concerns that the AI Act will lead to the end of innovation in marketing.
Source: SAP Emarsys
Shoppers more likely to trust user reviews than influencers
Nearly two-thirds (65%) of UK consumers say they have made a purchase based on online reviews or comments from fellow shoppers, as opposed to 58% who say they have made a purchase thanks to a social media endorsement.
Sports and leisure equipment (63%), decorative homewares (58%), luxury goods (56%), and cultural events (55%) are identified as product categories where consumers are most likely to find peer-to-peer information valuable.
Accurate product information was found to be a key factor in whether a review was positive or negative. Two-thirds (66%) of UK shoppers say that discrepancies between the product they receive and its description are a key reason for leaving negative reviews, whereas 40% of respondents say they have returned an item in the past year because the product details were inaccurate or misleading.
According to research by Akeeno, purchases driven by influencer activity have also declined since 2023, with 50% reporting having made a purchase based on influencer content in 2025 compared to 54% two years ago.
Source: Akeeno
77% of B2B marketing leaders say buyers still rely on their networks
When vetting what brands to work with, 77% of B2B marketing leaders say potential buyers still look at the company’s wider network as well as its own channels.
Given the amount of content professionals are faced with, they are more likely to rely on other professionals they already know and trust, according to research from LinkedIn.
More than two-fifths (43%) of B2B marketers globally say their network is still their primary source for advice at work, ahead of family and friends, search engines, and AI tools.
Additionally, younger professionals surveyed say they are still somewhat sceptical of AI, with three-quarters (75%) of 18- to 24-year-olds saying that even as AI becomes more advanced, there’s still no substitute for the intuition and insights they get from trusted colleagues.
Since professionals are more likely to trust content and advice from peers, marketers are now investing more in creators, employees, and subject matter experts to build trust. As a result, 80% of marketers say trusted creators are now essential to earning credibility with younger buyers.
Source: LinkedIn
Business confidence up 11 points but leaders remain concerned about economy
Business leader confidence has increased slightly from last month, having risen from -72 in July to -61 in August.
The IoD Directors’ Economic Confidence Index, which measures business leader optimism in prospects for the UK economy, is now back to where it was immediately after last year’s Budget.
This improvement comes from several factors, including the rise in investment intentions (up from -27 in July to -8 in August), the rise in headcount expectations from -23 to -4 over the same period, and the increase in revenue expectations from -8 to 12.
Additionally, business leaders’ confidence in their own organisations is also up, standing at 1 in August compared to -9 in July.
Several factors were identified as being of concern for business leaders; these include UK economic conditions at 76%, up from 67% in May, and both employment taxes (remaining at 59%) and business taxes (up to 47%, from 45%) continuing to be of significant concern.
Source: The Institute of Directors
Total volume of alcohol sold in retail down 2.3%
The total volume of alcohol sold in retail has fallen by 2.3% in the first half of 2025 compared to the previous year, equivalent to 90 million fewer litres. Value sales are also down by 1.1% compared to the same period in 2024.
At the same time, retail sales of non-alcoholic drinks have increased by 5.5% compared to last year, while volume sales are up by 2.3%, equivalent to a further 1.5 billion litres.
As the demand for non-alcoholic beverages grows, people increasingly expect these options to be available in their local bars and restaurants, with 55% of Brits and Europeans now expecting bars to always serve non-alcoholic beer.
As well as this, there are shifts happening within the alcoholic beverages category with value sales of no and low-alcohol spirits rising by 16.1%, and sales of ready-to-drink spirits growing by 11.6% compared to last year.
Source: Circana
Ethics & Policy
AI ethics under scrutiny, young people most exposed

New reports into the rise of artificial intelligence (AI) showed incidents linked to ethical breaches have more than doubled in just two years.
At the same time, entry-level job opportunities have been shrinking, partly due to the spread of this automation.
AI is moving from the margins to the mainstream at extraordinary speed and both workplaces and universities are struggling to keep up.
Tools such as ChatGPT, Gemini and Claude are now being used to draft emails, analyse data, write code, mark essays and even decide who gets a job interview.
Alongside this rapid rollout, a March report from McKinsey, one by the OECD in July and an earlier Rand report warned of a sharp increase in ethical controversies — from cheating scandals in exams to biased recruitment systems and cybersecurity threats — leaving regulators and institutions scrambling to respond.
The McKinsey survey said almost eight in 10 organisations now used AI in at least one business function, up from half in 2022.
While adoption promises faster workflows and lower costs, many companies deploy AI without clear policies. Universities face similar struggles, with students increasingly relying on AI for assignments and exams while academic rules remain inconsistent, it said.
The OECD’s AI Incidents and Hazards Monitor reported that ethical and operational issues involving AI have more than doubled since 2022.
Common concerns included accountability — who is responsible when AI errs; transparency — whether users understand AI decisions; and fairness, whether AI discriminates against certain groups.
Many models operated as “black boxes”, producing results without explanation, making errors hard to detect and correct, it said.
In workplaces, AI is used to screen CVs, rank applicants, and monitor performance. Yet studies show AI trained on historical data can replicate biases, unintentionally favouring certain groups.
Rand reported that AI was also used to manipulate information, influence decisions in sensitive sectors, and conduct cyberattacks.
Meanwhile, 41 per cent of professionals report that AI-driven change is harming their mental health, with younger workers feeling most anxious about job security.
LinkedIn data showed that entry-level roles in the US have fallen by more than 35 per cent since 2023, while 63 per cent of executives expected AI to replace tasks currently done by junior staff.
Aneesh Raman, LinkedIn’s chief economic opportunity officer, described this as “a perfect storm” for new graduates: Hiring freezes, economic uncertainty and AI disruption, as the BBC reported August 26.
LinkedIn forecasts that 70 per cent of jobs will look very different by 2030.
Recent Stanford research confirmed that employment among early-career workers in AI-exposed roles has dropped 13 per cent since generative AI became widespread, while more experienced workers or less AI-exposed roles remained stable.
Companies are adjusting through layoffs rather than pay cuts, squeezing younger workers out, it found.
In Belgium, AI ethics and fairness debates have intensified following a scandal in Flanders’ medical entrance exams.
Investigators caught three candidates using ChatGPT during the test.
Separately, 19 students filed appeals, suspecting others may have used AI unfairly after unusually high pass rates: Some 2,608 of 5,544 participants passed but only 1,741 could enter medical school. The success rate jumped to 47 per cent from 18.9 per cent in 2024, raising concerns about fairness and potential AI misuse.
Flemish education minister Zuhal Demir condemned the incidents, saying students who used AI had “cheated themselves, the university and society”.
Exam commission chair Professor Jan Eggermont noted that the higher pass rate might also reflect easier questions, which were deliberately simplified after the previous year’s exam proved excessively difficult, as well as the record number of participants, rather than AI-assisted cheating alone.
French-speaking universities, in the other part of the country, were not concerned by this scandal, as they still conduct medical entrance exams entirely on paper, something Demir said he was considering going back to.
Ethics & Policy
Governing AI with inclusion: An Egyptian model for the Global South

When artificial intelligence tools began spreading beyond technical circles and into the hands of everyday users, I saw a real opportunity to understand this profound transformation and harness AI’s potential to benefit Egypt as a state and its citizens. I also had questions: Is AI truly a national priority for Egypt? Do we need a legal framework to regulate it? Does it provide adequate protection for citizens? And is it safe enough for vulnerable groups like women and children?
These questions were not rhetorical. They were the drivers behind my decision to work on a legislative proposal for AI governance. My goal was to craft a national framework rooted in inclusion, dialogue, and development, one that does not simply follow global trends but actively shapes them to serve our society’s interests. The journey Egypt undertook can offer inspiration for other countries navigating the path toward fair and inclusive digital policies.
Egypt’s AI Development Journey
Over the past five years, Egypt has accelerated its commitment to AI as a pillar of its Egypt Vision 2030 for sustainable development. In May 2021, the government launched its first National AI Strategy, focusing on capacity building, integrating AI in the public sector, and fostering international collaboration. A National AI Council was established under the Ministry of Communications and Information Technology (MCIT) to oversee implementation. In January 2025, President Abdel Fattah El-Sisi unveiled the second National AI Strategy (2025–2030), which is built around six pillars: governance, technology, data, infrastructure, ecosystem development, and capacity building.
Since then, the MCIT has launched several initiatives, including training 100,000 young people through the “Our Future is Digital” programme, partnering with UNESCO to assess AI readiness, and integrating AI into health, education, and infrastructure projects. Today, Egypt hosts AI research centres, university departments, and partnerships with global tech companies—positioning itself as a regional innovation hub.
AI-led education reform
AI is not reserved for startups and hospitals. In May 2025, President El-Sisi instructed the government to consider introducing AI as a compulsory subject in pre-university education. In April 2025, I formally submitted a parliamentary request and another to the Deputy Prime Minister, suggesting that the government include AI education as part of a broader vision to prepare future generations, as outlined in Egypt’s initial AI strategy. The political leadership’s support for this proposal highlighted the value of synergy between decision-makers and civil society. The Ministries of Education and Communications are now exploring how to integrate AI concepts, ethics, and basic programming into school curricula.
From dialogue to legislation: My journey in AI policymaking
As Deputy Chair of the Foreign Affairs Committee in Parliament, I believe AI policymaking should not be confined to closed-door discussions. It must include all voices. In shaping Egypt’s AI policy, we brought together:
- The private sector, from startups to multinationals, will contribute its views on regulations, data protection, and innovation.
- Civil society – to emphasise ethical AI, algorithmic justice, and protection of vulnerable groups.
- International organisations, such as the OECD, UNDP, and UNESCO, share global best practices and experiences.
- Academic institutions – I co-hosted policy dialogues with the American University in Cairo and the American Chamber of Commerce (AmCham) to discuss governance standards and capacity development.
From recommendations to action: The government listening session
To transform dialogue into real policy, I formally requested the MCIT to host a listening session focused solely on the private sector. Over 70 companies and experts attended, sharing their recommendations directly with government officials.
This marked a key turning point, transitioning the initiative from a parliamentary effort into a participatory, cross-sectoral collaboration.
Drafting the law: Objectives, transparency, and risk-based classification
Based on these consultations, participants developed a legislative proposal grounded in transparency, fairness, and inclusivity. The proposed law includes the following core objectives:
- Support education and scientific research in the field of artificial intelligence
- Provide specific protection for individuals and groups most vulnerable to the potential risks of AI technologies
- Govern AI systems in alignment with Egypt’s international commitments and national legal framework
- Enhance Egypt’s position as a regional and international hub for AI innovation, in partnership with development institutions
- Support and encourage private sector investment in the field of AI, especially for startups and small enterprises
- Promote Egypt’s transition to a digital economy powered by advanced technologies and AI
To operationalise these objectives, the bill includes:
- Clear definitions of AI systems
- Data protection measures aligned with Egypt’s 2020 Personal Data Protection Law
- Mandatory algorithmic fairness, transparency, and auditability
- Incentives for innovation, such as AI incubators and R&D centres
Establishment of ethics committees and training programmes for public sector staff
The draft law also introduces a risk-based classification framework, aligning it with global best practices, which categorises AI systems into three tiers:
1. Prohibited AI systems – These are banned outright due to unacceptable risks, including harm to safety, rights, or public order.
2. High-risk AI systems – These require prior approval, detailed documentation, transparency, and ongoing regulatory oversight. Common examples include AI used in healthcare, law enforcement, critical infrastructure, and education.
3. Limited-risk AI systems – These are permitted with minimal safeguards, such as user transparency, labelling of AI-generated content, and optional user consent. Examples include recommendation engines and chatbots.
This classification system ensures proportionality in regulation, protecting the public interest without stifling innovation.
Global recognition: The IPU applauds Egypt’s model
The Inter-Parliamentary Union (IPU), representing over 179 national parliaments, praised Egypt’s AI bill as a model for inclusive AI governance. It highlighted that involving all stakeholders builds public trust in digital policy and reinforces the legitimacy of technology laws.
Key lessons learned
- Inclusion builds trust – Multistakeholder participation leads to more practical and sustainable policies.
- Political will matters – President El-Sisi’s support elevated AI from a tech topic to a national priority.
- Laws evolve through experience – Our draft legislation is designed to be updated as the field develops.
- Education is the ultimate infrastructure – Bridging the future digital divide begins in the classroom.
- Ethics come first – From the outset, we established values that focus on fairness, transparency, and non-discrimination.
Challenges ahead
As the draft bill progresses into final legislation and implementation, several challenges lie ahead:
- Training regulators on AI fundamentals
- Equipping public institutions to adopt ethical AI
- Reducing the urban-rural digital divide
- Ensuring national sovereignty over data
- Enhancing Egypt’s global role as a policymaker—not just a policy recipient
Ensuring representation in AI policy
As a female legislator leading this effort, it was important for me to prioritise the representation of women, youth, and marginalised groups in technology policymaking. If AI is built on biased data, it reproduces those biases. That’s why the policymaking table must be round, diverse, and representative.
A vision for the region
I look forward to seeing Egypt:
- Advance regional AI policy partnerships across the Middle East and Africa
- Embedd AI ethics in all levels of education
- Invest in AI for the public good
Because AI should serve people—not control them.
Better laws for a better future
This journey taught me that governing AI requires courage to legislate before all the answers are known—and humility to listen to every voice. Egypt’s experience isn’t just about technology; it’s about building trust and shared ownership. And perhaps that’s the most important infrastructure of all.
The post Governing AI with inclusion: An Egyptian model for the Global South appeared first on OECD.AI.
-
Business3 days ago
The Guardian view on Trump and the Fed: independence is no substitute for accountability | Editorial
-
Tools & Platforms3 weeks ago
Building Trust in Military AI Starts with Opening the Black Box – War on the Rocks
-
Ethics & Policy1 month ago
SDAIA Supports Saudi Arabia’s Leadership in Shaping Global AI Ethics, Policy, and Research – وكالة الأنباء السعودية
-
Events & Conferences3 months ago
Journey to 1000 models: Scaling Instagram’s recommendation system
-
Jobs & Careers2 months ago
Mumbai-based Perplexity Alternative Has 60k+ Users Without Funding
-
Funding & Business2 months ago
Kayak and Expedia race to build AI travel agents that turn social posts into itineraries
-
Education2 months ago
VEX Robotics launches AI-powered classroom robotics system
-
Podcasts & Talks2 months ago
Happy 4th of July! 🎆 Made with Veo 3 in Gemini
-
Podcasts & Talks2 months ago
OpenAI 🤝 @teamganassi
-
Mergers & Acquisitions2 months ago
Donald Trump suggests US government review subsidies to Elon Musk’s companies