2025人工智能状况报告_拨开炒作迷雾,_驾驭信任_安全与价值_61页_10mb

Summary of "The State of AI: Go Beyond the Hype to Navigate Trust, Security and Value"

Core Content

This report provides an in-depth analysis of the current state of AI adoption in enterprises, focusing on the challenges and opportunities related to trust, security, and value creation. It highlights the increasing use of generative AI tools, the need for stronger governance, and the importance of information management strategies to mitigate risks and maximize benefits.

Main Points

AI Adoption is Accelerating

• Generative AI tools like Microsoft 365 Copilot and ChatGPT are widely used, with Microsoft 365 Copilot showing faster growth.
• Over 50% of employees use generative AI tools in their daily or weekly work.
• 88.3% of organizations start with a pilot program to evaluate AI feasibility and risks, while 11.7% jump directly into production.

Data Security and Quality Concerns

• 75.1% of organizations experienced AI-related security breaches in the past year.
• 68.7% of organizations report that inaccurate AI output due to outdated or irrelevant data is the biggest concern.
• 68.5% cite data security concerns as a major barrier to AI rollout.
• 81.3% of organizations delayed AI deployment due to data security and management issues.

Governance and Policy Development

• 98% of organizations expect to have an AI Acceptable Use policy within the next 12 months.
• 84.5% of organizations are enforcing or developing clear AI usage guidelines.
• 93.2% of respondents believe that internal teams have the highest responsibility for developing AI guidelines, followed by industry coalitions and government.
• Government and industry coalitions are seen as more important in APAC and public sector organizations.

Employee and Shadow AI Use

• 18.7% of organizations that approve only one AI tool admit they don’t know what other tools employees are using.
• 15% lower shadow AI usage is observed in organizations that sanction two or more tools.
• 99.5% of organizations have implemented interventions to improve AI literacy among employees.

Implementation Challenges

• 40% of employees currently have access to corporate-approved AI assistants, with 54.6% expected to have access in 12 months.
• 64.4% of organizations plan to increase investment in third-party governance tools, followed by data security tools (54.5%).

Information Management Strategies

• 90.6% of organizations have an information management program or framework in place.
• 50.6% of respondents rate automated data classification as extremely mature, but only 30.3% are fully satisfied with their current data classification processes.
• 77.2% of organizations with high information management maturity experienced data security incidents, indicating a gap between perceived effectiveness and actual outcomes.

Key Findings

• Data security is the top reason for delayed AI rollouts.
• Shadow AI use is growing, highlighting the need for better governance.
• AI literacy is being actively addressed through various interventions.
• Third-party tools are becoming more essential for managing AI risks and ensuring compliance.
• Information management maturity is often overstated, especially in data classification and incident prevention.

Recommendations

• Strengthen AI governance and ensure all AI policies are backed by clear procedures and technical controls.
• Invest in third-party governance and security tools to address AI-related risks effectively.
• Improve data classification and ensure robust data security measures are in place.
• Enhance employee training and awareness to reduce shadow AI usage and improve AI adoption.
• Monitor and update AI policies continuously to keep pace with technological advancements and new risks.

Conclusion

The report underscores that while AI is increasingly adopted across enterprises, the focus on governance, security, and data management remains critical. Organizations are recognizing the need for proactive strategies to address the growing risks associated with AI, especially as data volumes and complexity increase. Success in AI implementation is now measured by impact rather than mere usage, and the path to achieving this requires a balanced approach between technical capabilities and human-centric strategies.