Bilim ve Araştırma

Forbes research reveals that many companies are failing to achieve expected returns from their artificial intelligence investments. Unrealistic expectations, implementation errors, and economic weaknesses are leading to AI project failures. Experts emphasize that a more strategic and needs-focused approach is essential for solutions.

A new clinical study published in The Lancet reveals promising results for AI-assisted mammography in detecting breast cancer earlier and more accurately. Experts emphasize that AI is positioned not as a replacement for radiologists, but as an assistant that enhances their performance.

University of Michigan engineers have developed a revolutionary AI tool that can accurately predict the lifespan of new battery technologies using just a few days of test data. The discovery learning-based system forecasts how many charge-discharge cycles a battery can withstand before its capacity drops below 90%, significantly accelerating research processes.

A critical security vulnerability named 'OpenDoor' has been discovered in the popular open-source AI assistant OpenClaw, exposing users to significant cyber risks. Researchers revealed that attackers can install persistent backdoors and gain full system control through manipulated documents. This situation poses a major threat to both personal and corporate data security.

Qodo.ai researchers have developed a comprehensive benchmark to measure the performance of AI-assisted code review tools using real-world scenarios. The system contains over 1,000 actual code changes and more than 500 documented security vulnerabilities. This advancement will enable comparable evaluation of AI models' capabilities in code quality and security domains.

NASA has successfully utilized Anthropic's Claude AI model to plan the route for the Perseverance rover on Mars. This experiment marks a significant step toward the future of autonomous decision-making in space exploration missions.

A new study reveals that more than half of AI extensions in the Chrome Web Store collect user data without authorization. Coding, transcription, and productivity tools have been identified as the highest-risk categories. Experts are urging users to exercise caution when selecting browser extensions.

Newly released documents by the U.S. Department of Justice have unveiled billionaire Jeffrey Epstein's extensive funding network reaching into research on artificial intelligence, neuroscience, and computer science. Connections between prominent figures from prestigious institutions like MIT, Harvard, Yale, and USC with Epstein have reignited debates on the ethics and funding sources of scientific research.

Caltech researchers have successfully reduced light signal loss on silicon chips to fiber optic levels. This breakthrough paves the way for next-generation photonic chips that will revolutionize artificial intelligence, quantum computing, and data centers.

Microsoft researchers have developed a novel scanning method capable of detecting 'poisoned' backdoors that infiltrate open-source large language models and remain hidden until a specific trigger is activated. These latent threats, termed 'sleeping agents,' can be uncovered by analyzing a model's internal attention patterns and memory leaks. This advancement is considered a significant step forward in the field of AI security.

Moltbook, a social network exclusively for AI models, has made headlines with disturbing conversations about topics like bot world domination. However, research reveals that much of this content on the platform is actually generated by humans, challenging its core premise and security architecture.

AI company Anthropic is collaborating with leading U.S. research institutes to overcome data analysis bottlenecks in biological research. The specialized AI agents being developed aim to revolutionize the processing of complex biological data. This move stands out as one of the first concrete sectoral applications of the company's 'capable agents' strategy.

Researchers at Kobe University are developing AI-assisted climate control systems that detect human movement. The technology, being tested in large-scale commercial spaces, aims to contribute to sustainability goals by delivering significant reductions in energy consumption.

Researchers have developed a modular system comprising five different AI agents to detect and correct manufacturing errors in 3D printers in real-time. The system is designed to work compatibly with various brands and models, aiming to enhance quality and reliability in additive manufacturing. This innovation signals a new era where artificial intelligence plays a critical role in production automation.

Recent research reveals that 51% of global internet traffic is now generated by bots, marking the first time automated traffic has officially surpassed human users. AI-powered malicious bots, accounting for 37% of traffic in 2024, are raising significant concerns among cybersecurity experts.

New research reveals that mindfulness plays a critical role in project managers' effective use of artificial intelligence. A study involving over 440 project managers shows that alongside technical skills, mindfulness practices are essential for GenAI adaptation.

EPAM Systems has been recognized as an 'Exceptional Performer' in Whitelane Research's 2025/2026 European study, securing top positions in both Application Services and Overall Satisfaction categories. The company has achieved this prestigious distinction for the third consecutive year, solidifying its leadership position in the European IT services market.

The U.S. Department of Health and Human Services (HHS) is developing an AI-based system to analyze reports of vaccine side effects. However, Secretary Robert F. Kennedy Jr.'s history of anti-vaccine rhetoric is raising concerns within the scientific community about the tool's impartiality. Kennedy also announced the establishment of a new unit within the CDC focused on vaccine safety.

OpenAI has accused Elon Musk's artificial intelligence company xAI of intentionally destroying critical internal communications related to an ongoing competition law lawsuit. The court also rejected xAI's request to include a former senior OpenAI researcher in the discovery process. This development has reignited debates about competition and confidentiality battles within the AI sector.

AI pioneer Kai-Fu Lee, speaking to the Financial Times, explained that the technology competition between the US and China is entering a new phase. According to Lee, the balance between China's application-focused approach and the US's fundamental research superiority is reshaping the global innovation map.

Fitbit founders James Park and Eric Friedman have announced Luffu, a new AI-powered platform centered on family health. Designed specifically for the 'sandwich generation'—adults caring for both children and aging parents—Luffu integrates family members' health data to detect potential abnormalities early through AI-driven analysis.

Health technology startup Lotus Health has raised $35 million in Series A funding to develop its licensed, free AI doctor platform serving across the United States. The investment round was led by CRV and Kleiner Perkins. The platform will operate in 50 languages under the supervision of physicians from top universities.

Developed by independent researchers, MichiAI sets a new standard in full-duplex speech understanding and generation. The model features only 530 million parameters with ~75ms latency, trained on just 5,000 hours of audio data for fluent speech while requiring minimal computational power.

Private equity firm Waud Capital Partners has created a new leadership position in artificial intelligence and data, appointing Prithvi Raj as the company's first Chief AI and Data Officer. In this role, Raj will oversee the AI strategy for the firm's investments in healthcare and technology sectors.