Encouraging deception in compaction summaries · OpenAI Alignment

Research on aligning AI with human values and intent, and reports documenting model failures.

saved by

• Uzay Girit