- Project Flux
- Posts
- AI in 2025: Safety Risks, Smarter Research & Project Disruption
AI in 2025: Safety Risks, Smarter Research & Project Disruption
From OpenAI’s ‘Deep Research’ to AI safety concerns and blockchain arbitration—explore how AI is reshaping industries, project delivery, and decision-making.
Proudly sponsored by ConstructAI, brought to you by Weston Analytics
This Week in Project Flux
This Week’s BIG thing
The AI landscape has reached an inflection point. The emergence of DeepSeek-R1 has triggered a new arms race, with major AI players—OpenAI, Microsoft, and Google DeepMind—rushing to accelerate development, seemingly at the expense of safety. Former OpenAI researchers have sounded the alarm, warning that AI safety is being sidelined in favour of market dominance.
The Fragmented AI Safety Landscape
Despite calls for responsible AI, the current approach is disjointed and ineffective. Here’s where things stand:
Major AI safety leaders are losing influence: OpenAI’s Superalignment team was disbanded in 2024, and former researchers describe the pace of development as "terrifying."
AI safety frameworks remain underfunded: RLHF is proving inadequate for advanced models, while more promising methods like Constitutional AI are years behind.
Regulations are too slow to keep up: The EU AI Act won’t take effect until 2026, the US has abandoned key AI safety orders, and China prioritises control over alignment.
Ethics vs. market forces is a losing battle: Companies are prioritising speed over safety, and the question of AI values remains unresolved.
Open-source AI is a double-edged sword: While increasing accessibility, it also makes powerful models available to bad actors.
The Harsh Reality: AI Safety is Losing
Three major challenges are preventing real progress:
The illusion of control: AI labs are not slowing down despite earlier assurances.
Geopolitical competition: AI is now a matter of national security, overriding safety concerns.
Lack of global oversight: Unlike nuclear weapons, AI has no enforceable international governance.
What Can Be Done?
☕️ Read the full story to get our take on what can be done to make AI safer.
What’s new: Tech
OpenAI has just shaken up the AI landscape with the release of o3-mini, a free advanced reasoning model now available to all ChatGPT users. This faster, more accurate model boosts capabilities in maths, coding, and scientific reasoning, bringing enterprise-grade AI to the masses—without a paywall.
Why This Matters
24% Faster & 39% Fewer Errors – A serious upgrade from its predecessor, o1-mini.
Smarter AI for Free – No cost barriers mean businesses and individuals can integrate AI seamlessly into their workflows.
Custom Reasoning Levels – Choose between low, medium, and high effort to balance speed and accuracy.
AI-Powered Project Delivery – From forecasting to automation, o3-mini streamlines tasks and improves decision-making.
What’s the Strategy?
OpenAI isn’t just making a model—it’s making a statement. While competitors like DeepSeek are pushing open-source alternatives, OpenAI is betting on performance, accessibility, and deep ecosystem integration.
How o3-mini Can Supercharge Your Projects
Automate Reports & Forecasting – No more spreadsheets; AI-powered insights in real time.
Smarter Risk Assessment – AI-driven insights help prevent costly mistakes before they happen.
AI That Talks to Your Tools – Potential integration with project software
Compliance Made Easy – AI ensures project documentation meets industry standards automatically.
The AI arms race is heating up, and 2025 is shaping up to be the year AI becomes an essential business tool—not just a luxury. Want to see o3-mini in action?
Check how it’s performing | Feature comparison with OpenAI’s o1 | Watch o3 in action | See how businesses are using it
AI is evolving fast—are you keeping up?
The Rabbit Hole
💙 Other tech news we’re loving
🚀 Meta on Edge Over DeepSeek’s AI Advancements - Meta’s AI teams are reportedly alarmed by DeepSeek’s rapid progress, with claims that OpenAI’s proprietary data may have fuelled its growth. Read more on Financial Times.
🔐 DeepSeek Data Breach Exposes Security Risks - DeepSeek’s viral AI model R1 faced a large-scale attack, forcing a halt on user registrations. A stark reminder of AI’s growing cybersecurity challenges. Details on The Verge.
📉 Nvidia’s Stock Takes a $500B Hit - The rise of DeepSeek’s R1 model is shaking up the AI market, with Nvidia losing $500B in value since its release. Explore the financial impact on BBC.
📱 DeepSeek Overtakes ChatGPT as the App Store’s #1 - In a major shift, DeepSeek has dethroned ChatGPT as the top AI app, highlighting a change in consumer preference. Read more on TechCrunch.
💰 SoftBank Eyes a $25B Investment in OpenAI - SoftBank is in talks to invest up to $25B in OpenAI, potentially its largest AI bet yet. What could this mean for the industry? More on Financial Times.
📞 Google’s AI Can Now Call Businesses for You - Google’s experimental “Ask for Me” feature allows its AI to book appointments on your behalf. The future of automated scheduling? Find out on Mashable.
🔊 ElevenLabs Secures Major Funding for AI Voice Tech - ElevenLabs has raised significant funding to expand its AI-powered voice technology, pushing the boundaries of synthetic speech. Full story coming soon.
Presented by Cogram
Cogram is an AI assistant specifically designed for AEC to help professionals save time on documentation and prevent disputes.
Automated meeting minutes: Cogram drafts meeting minutes directly in your company’s template with tailored levels of detail, format, and style.
Works for virtual and in-person meetings: Cogram works across Zoom, Teams, and in-person meetings, such as site inspections and kick-off meetings.
Field reports in minutes: Cogram’s mobile app combines site photos with minutes from on-site discussions to draft field reports in minutes.
Project-wide insights: Ask questions about your projects and get answers drawn from every relevant meeting and email chain from that project for the most up-to-date information.
What’s new: Projects
AI, Blockchain & the Future of Dispute Resolution in Construction
Can AI and blockchain revolutionise dispute resolution in construction? In a recent episode of Own The Build, Rob Dean, Associate Director at Diales, explored how technology is transforming arbitration. From AI-powered contract analysis to blockchain-based dispute resolution, the industry is set for a digital shake-up. Listen to the full episode here.
What Does This Mean for Me?
For project managers, QSs, and contract professionals, AI-driven arbitration could mean fewer costly delays and greater transparency. Smart contracts can automate compliance checks, while AI can analyse vast amounts of legal data in seconds. The result? Faster, fairer resolutions and reduced legal risks in construction projects.
Key Themes & Takeaways
Decentralised Arbitration: Platforms like Kleros introduce blockchain-based dispute resolution.
AI-Powered Legal Insights: AI assists in contract drafting, dispute prediction, and legal analysis.
Automation in QS & Contract Management: Routine admin tasks could be handled by AI, freeing professionals for strategic roles.
Lessons from Dubai Arbitration Week: Growing industry consensus on AI’s role in transforming dispute resolution.
Down the Rabbit Hole
As AI and blockchain gain traction, construction disputes may soon be settled with speed and precision. The question is—are you ready to adapt?
💙 Other Project news we’re loving
📊 Movar Launches AI-Powered Digital PMO for UK Utilities - Movar’s AI-driven Digital PMO promises to transform project management in the UK utilities sector, improving efficiency and decision-making. Learn more on PM Global.
🤖 Oracle Joins the AI Agent War - Oracle unveils new AI agents, entering the battle for enterprise AI dominance alongside OpenAI and Google. A game-changer for businesses? Full details on Yahoo Finance.
🦾 Figure AI Focuses on Workplace Robot Safety - Figure AI launches the Center for the Advancement of Humanoid Safety, aiming to ensure robots in workplaces operate safely. Read more on TechCrunch.
⚖️ US Copyright Office Rules AI Prompts Aren’t Copyrightable - The US Copyright Office declares AI-generated prompts lack copyright protection, igniting debates over intellectual property in the AI era. More on Newsroom LOC.
💰 Goldman Sachs Eyes DeepSeek AI for Finance - Goldman Sachs explores DeepSeek AI for coding and banking tasks, while Meta considers it for ads. Security screening is key before adoption. Details on QZ.
🎶 TemPolor: AI-Driven Music Platform for Creators - TemPolor enables creators to easily find and use AI-curated music for projects, simplifying sound selection for films, ads, and media.
🧬 Reid Hoffman’s AI Drug Discovery Venture - LinkedIn co-founder Reid Hoffman partners with scientists to launch Manas AI, an AI-driven platform revolutionising cancer drug discovery. More on Manas AI.
🎸 Paul McCartney Fights AI Music Rip-Offs - Paul McCartney urges stricter rules to protect artists from AI-generated copies of their work, defending creative integrity. Read his stance on BBC.
What’s new: Productivity
OpenAI Unveils ‘Deep Research’: AI-Powered Analysis in Minutes
OpenAI has launched ‘Deep Research’, an AI tool designed to autonomously conduct complex, multi-step internet research, delivering comprehensive reports in a fraction of the time it would take a human analyst. This development positions OpenAI competitively against emerging AI firms like China’s DeepSeek.
What Does This Mean for Me?
‘Deep Research’ offers the potential to streamline project planning and decision-making. By rapidly synthesising data from diverse sources, the tool can provide insights into market trends, regulatory changes, and technological advancements, enhancing productivity and project delivery.
Key Themes or Details
Autonomous Research: The tool independently gathers and analyses information from multiple online sources, producing fully cited reports.
Rapid Analysis: Completes tasks in minutes that would traditionally require hours of human effort.
Professional Applications: Aimed at sectors like finance, science, and engineering, with potential benefits for construction and real estate professionals.
Competitive Landscape: The launch responds to advancements by competitors such as DeepSeek, highlighting the dynamic nature of the AI industry.
Down the Rabbit Hole
By leveraging ‘Deep Research’, industry professionals can stay ahead of the curve, making informed decisions with greater efficiency.
💙 Other productivity news we’re loving
🧪 OpenAI Partners with US National Labs for Scientific Research - OpenAI’s O1 and O3 models will be available to 15,000 scientists at US National Labs, advancing nuclear research and AI-driven discoveries. More on TechCrunch.
🧠 Microsoft Makes Think Deeper Free for Copilot Users - Microsoft’s Think Deeper, its advanced reasoning model, is now free for all Copilot users, boosting AI-powered problem-solving and analysis. Read more on Microsoft.
🏛️ ChatGPT Gov: AI for US Government Agencies - OpenAI launches ChatGPT Gov, a tailored AI model designed to assist US government agencies with secure and efficient data processing. Learn more on OpenAI.
🎭 ChatGPT Introduces Customisable AI Traits - Users can now assign personality traits like ‘chatty’ or ‘Gen Z’ to ChatGPT, enhancing interactions and personalisation. Full details on TechCrunch.
Podcasts of the Week
🎙️ AI as Your Ultimate Thought Partner – A Conversation with Geoff Woods
In this episode of Project Flux, Geoff Woods, author of The AI-Driven Leader, explores how AI is transforming leadership, decision-making, and business strategy. He shares practical insights on leveraging AI as a thought partner, navigating industry disruptions, and the path to AGI. Listen now on Project Flux.
AI Ethics Showdown: OpenAI Faces Backlash Over Data Usage Claims
Giveaway
The Astute User’s Guide to AI: Practical Steps for Responsible Implementation
📖 Overview:
AI is reshaping industries, but using it effectively and ethically requires strategic thinking. This guide, developed by the Alan Turing Institute, is tailored for professionals providing actionable advice for making informed decisions about AI implementation.
🔎 What’s Inside:
✅ Key Questions to Ask: Ensure AI aligns with your goals.
✅ Avoid Common Pitfalls: Identify risks before they derail projects.
✅ Ethical & Regulatory Insights: Navigate AI’s evolving landscape responsibly.
✅ Real-World Examples: See how AI is transforming industries like yours.
📢 Download Now:
Access this expertly crafted resource by the Alan Turing Institute to deploy AI with confidence and clarity. Click here to download the full guide and share with your team to drive smarter, more responsible AI decisions.
Meet Project Flux: About Us
At Project Flux, we're committed to pioneering the future of construction and project delivery through the lens of cutting-edge Artificial Intelligence insights. Our vision is to be at the forefront of integrating AI into the fabric of project delivery, transforming how projects are conceptualised, planned, and executed.
Our Mission Project Flux aims to not only inform and educate but also to inspire professionals in the construction industry to embrace the transformative potential of AI. We believe in the power of AI to revolutionise project delivery, making it more efficient, predictive, and adaptable to the dynamic demands of the modern world.
What We Offer Through our insightful newsletters, podcasts and curated content on LinkedIn, and engaging discussions, Project Flux serves as a resource for professionals seeking to stay ahead in their field. We offer a blend of practical advice, thought leadership, and the latest developments in AI and construction technology.
What People Say…
“It was a real pleasure being a guest on the Project Flux podcast. James and Yoshi are really on top of things when it comes to AI in general and its application in project management specifically. If you just have a few minutes a week have a read through their newsletter so you can stay informed. If you have just a bit more time, they know how to ask the right questions in the podcast."